Corpus

We present a unique, extensive, and open synaptic physiology analysis platform and dataset. Through its application, we reveal principles that relate cell type to synaptic properties and intralaminar circuit organization in the mouse and human cortex. The dynamics of excitatory synapses align with the postsynaptic cell subclass, whereas inhibitory synapse dynamics partly align with presynaptic cell subclass but with considerable overlap. Synaptic properties are heterogeneous in most subclass-to-subclass connections. The two main axes of heterogeneity are strength and variability. Cell subclasses divide along the variability axis, whereas the strength axis accounts for substantial heterogeneity within the subclass. In the human cortex, excitatory-to-excitatory synaptic dynamics are distinct from those in the mouse cortex and vary with depth across layers 2 and 3.

. Here, using high-throughput transcriptomic and epigenomic profiling of more than 450,000 single nuclei in humans, marmoset monkeys and mice, we demonstrate a broadly conserved cellular makeup of this region, with similarities that mirror evolutionary distance and are consistent between the transcriptome and epigenome. The core conserved molecular identities of neuronal and non-neuronal cell types allow us to generate a cross-species consensus classification of cell types, and to infer conserved properties of cell types across species. Despite the overall conservation, however, many species-dependent specializations are apparent, including differences in cell-type proportions, gene expression, DNA methylation and chromatin state. Few cell-type marker genes are conserved across species, revealing a short list of candidate genes and regulatory mechanisms that are responsible for conserved features of homologous cell types, such as the GABAergic chandelier cells. This consensus transcriptomic classification allows us to use patch-seq (a combination of whole-cell patch-clamp recordings, RNA sequencing and morphological characterization) to identify corticospinal Betz cells from layer 5 in non-human primates and humans, and to characterize their highly specialized physiology and anatomy. These findings highlight the robust molecular underpinnings of cell-type diversity in M1 across mammals, and point to the genes and regulatory pathways responsible for the functional identity of cell types and their species-specific adaptations.

, yet our knowledge of its diversity remains limited. Here, to systematically examine complete single-neuron morphologies on a brain-wide scale, we established a pipeline encompassing sparse labelling, whole-brain imaging, reconstruction, registration and analysis. We fully reconstructed 1,741 neurons from cortex, claustrum, thalamus, striatum and other brain regions in mice. We identified 11 major projection neuron types with distinct morphological features and corresponding transcriptomic identities. Extensive projectional diversity was found within each of these major types, on the basis of which some types were clustered into more refined subtypes. This diversity follows a set of generalizable principles that govern long-range axonal projections at different levels, including molecular correspondence, divergent or convergent projection, axon termination pattern, regional specificity, topography, and individual cell variability. Although clear concordance with transcriptomic profiles is evident at the level of major projection type, fine-grained morphological diversity often does not readily correlate with transcriptomic subtypes derived from unsupervised clustering, highlighting the need for single-cell cross-modality studies. Overall, our study demonstrates the crucial need for quantitative description of complete single-cell anatomy in cell-type classification, as single-cell morphological diversity reveals a plethora of ways in which different cell types and their individual members may contribute to the configuration and function of their respective circuits.

Championing open science, an adversarial collaboration aims to unravel the footprints of consciousness

Fluorescent calcium indicators are often used to investigate neural dynamics, but the relationship between fluorescence and action potentials (APs) remains unclear. Most APs can be detected when the soma almost fills the microscope's field of view, but calcium indicators are used to image populations of neurons, necessitating a large field of view, generating fewer photons per neuron, and compromising AP detection. Here, we characterized the AP-fluorescence transfer function in vivo for 48 layer 2/3 pyramidal neurons in primary visual cortex, with simultaneous calcium imaging and cell-attached recordings from transgenic mice expressing GCaMP6s or GCaMP6f. While most APs were detected under optimal conditions, under conditions typical of population imaging studies, only a minority of 1 AP and 2 AP events were detected (often <10% and ~20-30%, respectively), emphasizing the limits of AP detection under more realistic imaging conditions.

Different neuron types serve distinct roles in neural processing. Extracellular electrical recordings are extensively used to study brain function but are typically blind to cell identity. Morphoelectrical properties of neurons measured on spatially dense electrode arrays have the potential to distinguish neuron types. We used high-density silicon probes to record from cortical and subcortical regions of the mouse brain. Extracellular waveforms of each neuron were detected across many channels and showed distinct spatiotemporal profiles among brain regions. Classification of neurons by brain region was improved with multichannel compared with single-channel waveforms. In visual cortex, unsupervised clustering identified the canonical regular-spiking (RS) and fast-spiking (FS) classes but also indicated a subclass of RS units with unidirectional backpropagating action potentials (BAPs). Moreover, BAPs were observed in many hippocampal RS cells. Overall, waveform analysis of spikes from high-density probes aids neuron identification and can reveal dendritic backpropagation. NEW &amp; NOTEWORTHY It is challenging to identify neuron types with extracellular electrophysiology in vivo. We show that spatiotemporal action potentials measured on high-density electrode arrays can capture cell type-specific morphoelectrical properties, allowing classification of neurons across brain structures and within the cortex. Moreover, backpropagating action potentials are reliably detected in vivo from subpopulations of cortical and hippocampal neurons. Together, these results enhance the utility of dense extracellular electrophysiology for cell-type interrogation of brain network function.

A thought-provoking argument that consciousness—more widespread than previously assumed—is the feeling of being alive, not a type of computation or a clever hack. In The Feeling of Life Itself, Christof Koch offers a straightforward definition of consciousness as any subjective experience, from the most mundane to the most exalted—the feeling of being alive. Psychologists study which cognitive operations underpin a given conscious perception. Neuroscientists track the neural correlates of consciousness in the brain, the organ of the mind. But why the brain and not, say, the liver? How can the brain—three pounds of highly excitable matter, a piece of furniture in the universe, subject to the same laws of physics as any other piece—give rise to subjective experience? Koch argues that what is needed to answer these questions is a quantitative theory that starts with experience and proceeds to the brain. In The Feeling of Life Itself, Koch outlines such a theory, based on integrated information. Koch describes how the theory explains many facts about the neurology of consciousness and how it has been used to build a clinically useful consciousness meter. The theory predicts that many, and perhaps all, animals experience the sights and sounds of life; consciousness is much more widespread than conventionally assumed. Contrary to received wisdom, however, Koch argues that programmable computers will not have consciousness. Even a perfect software model of the brain is not conscious. Its simulation is fake consciousness. Consciousness is not a special type of computation—it is not a clever hack. Consciousness is about being.

Electrophysiological methods are the gold standard in neuroscience because they reveal the activity of individual neurons at high temporal resolution and in arbitrary brain locations. Microelectrode arrays based on complementary metal-oxide semiconductor (CMOS) technology, such as Neuropixels probes, look set to transform these methods. Neuropixels probes provide ∼1000 recording sites on an extremely narrow shank, with on-board amplification, digitization, and multiplexing. They deliver low-noise recordings from hundreds of neurons, providing a step change in the type of data available to neuroscientists. Here we discuss the opportunities afforded by these probes for large-scale electrophysiology, the challenges associated with data processing and anatomical localization, and avenues for further improvements of the technology.

There is a high diversity of neuronal types in the mammalian neocortex. To facilitate construction of system models with multiple cell types, we generate a database of point models associated with the Allen Cell Types Database. We construct a set of generalized leaky integrate-and-fire (GLIF) models of increasing complexity to reproduce the spiking behaviors of 645 recorded neurons from 16 transgenic lines. The more complex models have an increased capacity to predict spiking behavior of hold-out stimuli. We use unsupervised methods to classify cell types, and find that high level GLIF model parameters are able to differentiate transgenic lines comparable to electrophysiological features. The more complex model parameters also have an increased ability to differentiate between transgenic lines. Thus, creating simple models is an effective dimensionality reduction technique that enables the differentiation of cell types from electrophysiological responses without the need for a priori-defined features. This database will provide a set of simplified models of multiple cell types for the community to use in network models.

The cellular components of mammalian neocortical circuits are diverse, and capturing this diversity in computational models is challenging. Here we report an approach for generating biophysically detailed models of 170 individual neurons in the Allen Cell Types Database to link the systematic experimental characterization of cell types to the construction of cortical models. We build models from 3D morphologies and somatic electrophysiological responses measured in the same cells. Densities of active somatic conductances and additional parameters are optimized with a genetic algorithm to match electrophysiological features. We evaluate the models by applying additional stimuli and comparing model responses to experimental data. Applying this technique across a diverse set of neurons from adult mouse primary visual cortex, we verify that models preserve the distinctiveness of intrinsic properties between subsets of cells observed in experiments. The optimized models are accessible online alongside the experimental data. Code for optimization and simulation is also openly distributed.

The powerful suite of available genetic tools is driving tremendous progress in understanding mouse brain cell types and circuits. However, the degree of conservation in human remains largely unknown in large part due to the lack of such tools and healthy tissue preparations. To close this gap, we describe a robust and stable adult human neurosurgically-derived ex vivo acute and cultured neocortical brain slice system optimized for rapid molecular-genetic manipulation. Surprisingly, acute human brain slices exhibited exceptional viability, and neuronal intrinsic membrane properties could be assayed for at least three days. Maintaining adult human slices in culture under sterile conditions further enabled the application of viral tools to drive rapid expression of exogenous transgenes. Widespread neuron-specific labeling was achieved as early as two days post infection with HSV-1 vectors, with virally-transduced neurons exhibiting membrane properties largely comparable to uninfected neurons over this short timeframe. Finally, we demonstrate the suitability of this culture paradigm for optical manipulation and monitoring of neuronal activity using genetically encoded probes, opening a path for applying modern molecular-genetic tools to study human brain circuit function.

Summary To understand how the brain processes sensory information to guide behavior, we must know how stimulus representations are transformed throughout the visual cortex. Here we report an open, large-scale physiological survey of neural activity in the awake mouse visual cortex: the Allen Brain Observatory Visual Coding dataset. This publicly available dataset includes cortical activity from nearly 60,000 neurons collected from 6 visual areas, 4 layers, and 12 transgenic mouse lines from 221 adult mice, in response to a systematic set of visual stimuli. Using this dataset, we reveal functional differences across these dimensions and show that visual cortical responses are sparse but correlated. Surprisingly, responses to different stimuli are largely independent, e.g. whether a neuron responds to natural scenes provides no information about whether it responds to natural movies or to gratings. We show that these phenomena cannot be explained by standard local filter-based models, but are consistent with multi-layer hierarchical computation, as found in deeper layers of standard convolutional neural networks.

This protocol is a practical guide to the N-methyl-D-glucamine (NMDG) protective recovery method of brain slice preparation. Numerous recent studies have validated the utility of this method for enhancing neuronal preservation and overall brain slice viability. The implementation of this technique by early adopters has facilitated detailed investigations into brain function using diverse experimental applications and spanning a wide range of animal ages, brain regions, and cell types. Steps are outlined for carrying out the protective recovery brain slice technique using an optimized NMDG artificial cerebrospinal fluid (aCSF) media formulation and enhanced procedure to reliably obtain healthy brain slices for patch clamp electrophysiology. With this updated approach, a substantial improvement is observed in the speed and reliability of gigaohm seal formation during targeted patch clamp recording experiments while maintaining excellent neuronal preservation, thereby facilitating challenging experimental applications. Representative results are provided from multi-neuron patch clamp recording experiments to assay synaptic connectivity in neocortical brain slices prepared from young adult transgenic mice and mature adult human neurosurgical specimens. Furthermore, the optimized NMDG protective recovery method of brain slicing is compatible with both juvenile and adult animals, thus resolving a limitation of the original methodology. In summary, a single media formulation and brain slicing procedure can be implemented across various species and ages to achieve excellent viability and tissue preservation.

Generating a comprehensive description of cortical networks requires a large-scale, systematic approach. To that end, we have begun a pipeline project using multipatch electrophysiology, supplemented with two-photon optogenetics, to characterize connectivity and synaptic signaling between classes of neurons in adult mouse primary visual cortex (V1) and human cortex. We focus on producing results detailed enough for the generation of computational models and enabling comparison with future studies. Here, we report our examination of intralaminar connectivity within each of several classes of excitatory neurons. We find that connections are sparse but present among all excitatory cell classes and layers we sampled, and that most mouse synapses exhibited short-term depression with similar dynamics. Synaptic signaling between a subset of layer 2/3 neurons, however, exhibited facilitation. These results contribute to a body of evidence describing recurrent excitatory connectivity as a conserved feature of cortical microcircuits.

ABSTRACT Neocortex contains a multitude of cell types segregated into layers and functionally distinct regions. To investigate the diversity of cell types across the mouse neocortex, we analyzed 12,714 cells from the primary visual cortex (VISp), and 9,035 cells from the anterior lateral motor cortex (ALM) by deep single-cell RNA-sequencing (scRNA-seq), identifying 116 transcriptomic cell types. These two regions represent distant poles of the neocortex and perform distinct functions. We define 50 inhibitory transcriptomic cell types, all of which are shared across both cortical regions. In contrast, 49 of 52 excitatory transcriptomic types were found in either VISp or ALM, with only three present in both. By combining single cell RNA-seq and retrograde labeling, we demonstrate correspondence between excitatory transcriptomic types and their region-specific long-range target specificity. This study establishes a combined transcriptomic and projectional taxonomy of cortical cell types from functionally distinct regions of the mouse cortex.

The role of the frontal cortex in consciousness remains a matter of debate. In this Perspective, we will critically review the clinical and neuroimaging evidence for the involvement of the front versus the back of the cortex in specifying conscious contents and discuss promising research avenues. Dual Perspectives Companion Paper: Should a Few Null Findings Falsify Prefrontal Theories of Conscious Perception?, by Brian Odegaard, Robert T. Knight, and Hakwan Lau

The connections between the claustrum and the cortex in mouse are systematically investigated with adeno-associated virus (AAV), an anterograde viral tracer. We first define the boundary and the three-dimensional structure of the claustrum based on a variety of molecular and anatomical data. From AAV injections into 42 neocortical and allocortical areas, we conclude that most cortical areas send bilateral projections to the claustrum, the majority being denser on the ipsilateral side. This includes prelimbic, infralimbic, medial, ventrolateral and lateral orbital, ventral retrosplenial, dorsal and posterior agranular insular, visceral, temporal association, dorsal and ventral auditory, ectorhinal, perirhinal, lateral entorhinal, and anteromedial, posteromedial, lateroposterior, laterointermediate, and postrhinal visual areas. In contrast, the cingulate and the secondary motor areas send denser projections to the contralateral claustrum than to the ipsilateral one. The gustatory, primary auditory, primary visual, rostrolateral visual, and medial entorhinal cortices send projections only to the ipsilateral claustrum. Primary motor, primary somatosensory and subicular areas barely send projections to either ipsi- or contralateral claustrum. Corticoclaustral projections are organized in a rough topographic manner, with variable projection strengths. We find that the claustrum, in turn, sends widespread projections preferentially to ipsilateral cortical areas with different projection strengths and laminar distribution patterns and to certain contralateral cortical areas. Our quantitative results show that the claustrum has strong reciprocal and bilateral connections with prefrontal and cingulate areas as well as strong reciprocal connections with the ipsilateral temporal and retrohippocampal areas, suggesting that it may play a crucial role in a variety of cognitive processes. J. Comp. Neurol. 525:1317-1346, 2017. © 2016 Wiley Periodicals, Inc.

Detailed anatomical understanding of the human brain is essential for unraveling its functional architecture, yet current reference atlases have major limitations such as lack of whole-brain coverage, relatively low image resolution, and sparse structural annotation. We present the first digital human brain atlas to incorporate neuroimaging, high-resolution histology, and chemoarchitecture across a complete adult female brain, consisting of magnetic resonance imaging (MRI), diffusion-weighted imaging (DWI), and 1,356 large-format cellular resolution (1 µm/pixel) Nissl and immunohistochemistry anatomical plates. The atlas is comprehensively annotated for 862 structures, including 117 white matter tracts and several novel cyto- and chemoarchitecturally defined structures, and these annotations were transferred onto the matching MRI dataset. Neocortical delineations were done for sulci, gyri, and modified Brodmann areas to link macroscopic anatomical and microscopic cytoarchitectural parcellations. Correlated neuroimaging and histological structural delineation allowed fine feature identification in MRI data and subsequent structural identification in MRI data from other brains. This interactive online digital atlas is integrated with existing Allen Institute for Brain Science gene expression atlases and is publicly accessible as a resource for the neuroscience community. J. Comp. Neurol. 524:3127-3481, 2016. © 2016 The Authors The Journal of Comparative Neurology Published by Wiley Periodicals, Inc.

The science of consciousness has made great strides by focusing on the behavioural and neuronal correlates of experience. However, while such correlates are important for progress to occur, they are not enough if we are to understand even basic facts, for example, why the cerebral cortex gives rise to consciousness but the cerebellum does not, though it has even more neurons and appears to be just as complicated. Moreover, correlates are of little help in many instances where we would like to know if consciousness is present: patients with a few remaining islands of functioning cortex, preterm infants, non-mammalian species and machines that are rapidly outperforming people at driving, recognizing faces and objects, and answering difficult questions. To address these issues, we need not only more data but also a theory of consciousness-one that says what experience is and what type of physical systems can have it. Integrated information theory (IIT) does so by starting from experience itself via five phenomenological axioms: intrinsic existence, composition, information, integration and exclusion. From these it derives five postulates about the properties required of physical mechanisms to support consciousness. The theory provides a principled account of both the quantity and the quality of an individual experience (a quale), and a calculus to evaluate whether or not a particular physical system is conscious and of what. Moreover, IIT can explain a range of clinical and laboratory findings, makes a number of testable predictions and extrapolates to a number of problematic conditions. The theory holds that consciousness is a fundamental property possessed by physical systems having specific causal properties. It predicts that consciousness is graded, is common among biological organisms and can occur in some very simple systems. Conversely, it predicts that feed-forward networks, even complex ones, are not conscious, nor are aggregates such as groups of individuals or heaps of sand. Also, in sharp contrast to widespread functionalist beliefs, IIT implies that digital computers, even if their behaviour were to be functionally equivalent to ours, and even if they were to run faithful simulations of the human brain, would experience next to nothing.

L5 pyramidal neurons are the only neocortical cell type with dendrites reaching all six layers of cortex, casting them as one of the main integrators in the cortical column. What is the nature and mode of computation performed in mouse primary visual cortex (V1) given the physiology of L5 pyramidal neurons? First, we experimentally establish active properties of the dendrites of L5 pyramidal neurons of mouse V1 using patch-clamp recordings. Using a detailed multi-compartmental model, we show this physiological setup to be well suited for coincidence detection between basal and apical tuft inputs by controlling the frequency of spike output. We further show how direct inhibition of calcium channels in the dendrites modulates such coincidence detection. To establish the singe-cell computation that this biophysics supports, we show that the combination of frequency-modulation of somatic output by tuft input and (simulated) calcium-channel blockage functionally acts as a composite sigmoidal function. Finally, we explore how this computation provides a mechanism whereby dendritic spiking contributes to orientation tuning in pyramidal neurons.

In this paper we provide an extensive evaluation of fixation prediction and salient object segmentation algorithms as well as statistics of major datasets. Our analysis identifies serious design flaws of existing salient object benchmarks, called the dataset design bias, by over emphasising the stereotypical concepts of saliency. The dataset design bias does not only create the discomforting disconnection between fixations and salient object segmentation, but also misleads the algorithm designing. Based on our analysis, we propose a new high quality dataset that offers both fixation and salient object segmentation ground-truth. With fixations and salient object being presented simultaneously, we are able to bridge the gap between fixations and salient objects, and propose a novel method for salient object segmentation. Finally, we report significant benchmark progress on 3 existing datasets of segmenting salient objects.

Multisensory integration is thought to require conscious perception. Although previous studies have shown that an invisible stimulus could be integrated with an audible one, none have demonstrated integration of two subliminal stimuli of different modalities. Here, pairs of identical or different audiovisual target letters (the sound /b/ with the written letter "b" or "m," respectively) were preceded by pairs of masked identical or different audiovisual prime digits (the sound /6/ with the written digit "6" or "8," respectively). In three experiments, awareness of the audiovisual digit primes was manipulated, such that participants were either unaware of the visual digit, the auditory digit, or both. Priming of the semantic relations between the auditory and visual digits was found in all experiments. Moreover, a further experiment showed that unconscious multisensory integration was not obtained when participants did not undergo prior conscious training of the task. This suggests that following conscious learning, unconscious processing suffices for multisensory integration.

Many decisions we make require visually identifying and evaluating numerous alternatives quickly. These usually vary in reward, or value, and in low-level visual properties, such as saliency. Both saliency and value influence the final decision. In particular, saliency affects fixation locations and durations, which are predictive of choices. However, it is unknown how saliency propagates to the final decision. Moreover, the relative influence of saliency and value is unclear. Here we address these questions with an integrated model that combines a perceptual decision process about where and when to look with an economic decision process about what to choose. The perceptual decision process is modeled as a drift-diffusion model (DDM) process for each alternative. Using psychophysical data from a multiple-alternative, forced-choice task, in which subjects have to pick one food item from a crowded display via eye movements, we test four models where each DDM process is driven by (i) saliency or (ii) value alone or (iii) an additive or (iv) a multiplicative combination of both. We find that models including both saliency and value weighted in a one-third to two-thirds ratio (saliency-to-value) significantly outperform models based on either quantity alone. These eye fixation patterns modulate an economic decision process, also described as a DDM process driven by value. Our combined model quantitatively explains fixation patterns and choices with similar or better accuracy than previous models, suggesting that visual saliency has a smaller, but significant, influence than value and that saliency affects choices indirectly through perceptual decisions that modulate economic decisions.

When monitoring neural activity using intracranial electrical recordings, researchers typically consider the signals to have two primary components: fast action potentials (APs) from neurons near the electrode, and the slower local field potential (LFP), thought to be dominated by postsynaptic currents integrated over a larger volume of tissue. In general, a decrease in signal power with increasing frequency is observed for most brain rhythms. The 100-200 Hz oscillations in the rat hippocampus, including "fast gamma" or "epsilon" oscillations and sharp wave-ripples (SPW-Rs), are one exception, showing an increase in power with frequency within this band. We have used detailed biophysical modeling to investigate the composition of extracellular potentials during fast oscillations in rat CA1. We find that postsynaptic currents exhibit a decreasing ability to generate large-amplitude oscillatory signals at high frequencies, whereas phase-modulated spiking shows the opposite trend. Our estimates indicate that APs and postsynaptic currents contribute similar proportions of the power contained in 140-200 Hz ripples, and the two combined generate a signal that closely resembles in vivo SPW-Rs. Much of the AP-generated signal originates from neurons further than 100 μm from the recording site, consistent with ripples appearing similarly strong regardless of whether or not they contain recognizable APs. Additionally, substantial power can be generated in the 90-150 Hz epsilon band by the APs of rhythmically firing pyramidal neurons. Thus, high-frequency LFPs may generally contain signatures of local cell assembly activation.

In which a scientist searches for an empirical explanation for phenomenal experience, spurred by his instinctual belief that life is meaningful.What links conscious experience of pain, joy, color, and smell to bioelectrical activity in the brain? How can anything physical give rise to nonphysical, subjective, conscious states? Christof Koch has devoted much of his career to bridging the seemingly unbridgeable gap between the physics of the brain and phenomenal experience. This engaging book—part scientific overview, part memoir, part futurist speculation—describes Koch's search for an empirical explanation for consciousness. Koch recounts not only the birth of the modern science of consciousness but also the subterranean motivation for his quest—his instinctual (if "romantic") belief that life is meaningful.Koch describes his own groundbreaking work with Francis Crick in the 1990s and 2000s and the gradual emergence of consciousness (once considered a "fringy" subject) as a legitimate topic for scientific investigation. Present at this paradigm shift were Koch and a handful of colleagues, including Ned Block, David Chalmers, Stanislas Dehaene, Giulio Tononi, Wolf Singer, and others. Aiding and abetting it were new techniques to listen in on the activity of individual nerve cells, clinical studies, and brain-imaging technologies that allowed safe and noninvasive study of the human brain in action. Koch gives us stories from the front lines of modern research into the neurobiology of consciousness as well as his own reflections on a variety of topics, including the distinction between attention and awareness, the unconscious, how neurons respond to Homer Simpson, the physics and biology of free will, dogs, Der Ring des Nibelungen, sentient machines, the loss of his belief in a personal God, and sadness. All of them are signposts in the pursuit of his life's work—to uncover the roots of consciousness.

What links conscious experience of pain, joy, color, and smell to bioelectrical activity in the brain? How can anything physical give rise to nonphysical, subjective, conscious states? Christof Koch has devoted much of his career to bridging the seemingly unbridgeable gap between the physics of the brain and phenomenal experience. This engaging book -- part scientific overview, part memoir, part futurist speculation -- describes Koch's search for an empirical explanation for consciousness. Koch recounts not only the birth of the modern science of consciousness but also the subterranean motivation for his quest -- his instinctual (if romantic) belief that life is meaningful. Koch describes his own groundbreaking work with Francis Crick in the 1990s and 2000s and the gradual emergence of consciousness (once considered a fringy subject) as a legitimate topic for scientific investigation. Present at this paradigm shift were Koch and a handful of colleagues, including Ned Block, David Chalmers, Stanislas Dehaene, Giulio Tononi, Wolf Singer, and others. Aiding and abetting it were new techniques to listen in on the activity of individual nerve cells, clinical studies, and brain-imaging technologies that allowed safe and noninvasive study of the human brain in action. Koch gives us stories from the front lines of modern research into the neurobiology of consciousness as well as his own reflections on a variety of topics, including the distinction between attention and awareness, the unconscious, how neurons respond to Homer Simpson, the physics and biology of free will, dogs, Der Ring des Nibelungen, sentient machines, the loss of his belief in a personal God, and sadness. All of them are signposts in the pursuit of his life's work -- to uncover the roots of consciousness.

Abstract Consumers often need to make very rapid choices among multiple brands (e.g., at a supermarket shelf) that differ both in their reward value (e.g., taste) and in their visual properties (e.g., color and brightness of the packaging). Since the visual properties of stimuli are known to influence visual attention, and attention is known to influence choices, this gives rise to a potential visual saliency bias in choices. We utilize experimental design from visual neuroscience in three real food choice experiments to measure the size of the visual saliency bias and how it changes with decision speed and cognitive load. Our results show that at rapid decision speeds visual saliency influences choices more than preferences do, that the bias increases with cognitive load, and that it is particularly strong when individuals do not have strong preferences among the options.

Detecting text regions in natural scenes is an important part of computer vision. We propose a novel text detection algorithm that extracts six different classes features of text, and uses Modest AdaBoost with multi-scale sequential search. Experiments show that our algorithm can detect text regions with a f= 0.70, from the ICDAR 2003 datasets which include images with text of various fonts, sizes, colors, alphabets and scripts.

We introduce a simple image descriptor referred to as the image signature. We show, within the theoretical framework of sparse signal mixing, that this quantity spatially approximates the foreground of an image. We experimentally investigate whether this approximate foreground overlaps with visually conspicuous image locations by developing a saliency algorithm based on the image signature. This saliency algorithm predicts human fixation points best among competitors on the Bruce and Tsotsos [1] benchmark data set and does so in much shorter running time. In a related experiment, we demonstrate with a change blindness data set that the distance between images induced by the image signature is closer to human perceptual distance than can be achieved using other saliency algorithms, pixel-wise, or GIST [2] descriptor methods.

Inspired by the primate visual system, computational saliency models decompose visual input into a set of feature maps across spatial scales in a number of pre-specified channels. The outputs of these feature maps are summed to yield the final saliency map. Here we use a least square technique to learn the weights associated with these maps from subjects freely fixating natural scenes drawn from four recent eye-tracking data sets. Depending on the data set, the weights can be quite different, with the face and orientation channels usually more important than color and intensity channels. Inter-subject differences are negligible. We also model a bias toward fixating at the center of images and consider both time-varying and constant factors that contribute to this bias. To compensate for the inadequacy of the standard method to judge performance (area under the ROC curve), we use two other metrics to comprehensively assess performance. Although our model retains the basic structure of the standard saliency model, it outperforms several state-of-the-art saliency algorithms. Furthermore, the simple structure makes the results applicable to numerous studies in psychophysics and physiology and leads to an extremely easy implementation for real-world applications.

One of the hallmarks of biological organisms is their ability to integrate disparate information sources to optimize their behavior in complex environments. How this capability can be quantified and related to the functional complexity of an organism remains a challenging problem, in particular since organismal functional complexity is not well-defined. We present here several candidate measures that quantify information and integration, and study their dependence on fitness as an artificial agent ("animat") evolves over thousands of generations to solve a navigation task in a simple, simulated environment. We compare the ability of these measures to predict high fitness with more conventional information-theoretic processing measures. As the animat adapts by increasing its "fit" to the world, information integration and processing increase commensurately along the evolutionary line of descent. We suggest that the correlation of fitness with information integration and with processing measures implies that high fitness requires both information processing as well as integration, but that information integration may be a better measure when the task requires memory. A correlation of measures of information integration (but also information processing) and fitness strongly suggests that these measures reflect the functional complexity of the animat, and that such measures can be used to quantify functional complexity even in the absence of fitness data.

Abstract An important open problem is how values are compared to make simple choices. A natural hypothesis is that the brain carries out the computations associated with the value comparisons in a manner consistent with the Drift Diffusion Model (DDM), since this model has been able to account for a large amount of data in other domains. We investigated the ability of four different versions of the DDM to explain the data in a real binary food choice task under conditions of high and low time pressure. We found that a seven-parameter version of the DDM can account for the choice and reaction time data with high-accuracy, in both the high and low time pressure conditions. The changes associated with the introduction of time pressure could be traced to changes in two key model parameters: the barrier height and the noise in the slope of the drift process.

The ability to choose rapidly among multiple targets embedded in a complex perceptual environment is key to survival. Targets may differ in their reward value as well as in their low-level perceptual properties (e.g., visual saliency). Previous studies investigated separately the impact of either value or saliency on choice; thus, it is not known how the brain combines these two variables during decision making. We addressed this question with three experiments in which human subjects attempted to maximize their monetary earnings by rapidly choosing items from a brief display. Each display contained several worthless items (distractors) as well as two targets, whose value and saliency were varied systematically. We compared the behavioral data with the predictions of three computational models assuming that (i) subjects seek the most valuable item in the display, (ii) subjects seek the most easily detectable item, and (iii) subjects behave as an ideal Bayesian observer who combines both factors to maximize the expected reward within each trial. Regardless of the type of motor response used to express the choices, we find that decisions are influenced by both value and feature-contrast in a way that is consistent with the ideal Bayesian observer, even when the targets' feature-contrast is varied unpredictably between trials. This suggests that individuals are able to harvest rewards optimally and dynamically under time pressure while seeking multiple targets embedded in perceptual clutter.

The brain's ability to handle sensory information is influenced by both selective attention and consciousness. There is no consensus on the exact relationship between these two processes and whether they are distinct. So far, no experiment has simultaneously manipulated both. We carried out a full factorial 2 x 2 study of the simultaneous influences of attention and consciousness (as assayed by visibility) on perception, correcting for possible concurrent changes in attention and consciousness. We investigated the duration of afterimages for all four combinations of high versus low attention and visible versus invisible. We show that selective attention and visual consciousness have opposite effects: paying attention to the grating decreases the duration of its afterimage, whereas consciously seeing the grating increases the afterimage duration. These findings provide clear evidence for distinctive influences of selective attention and consciousness on visual perception.

The cooperative action of neurons and glia generates electrical fields, but their effect on individual neurons via ephaptic interactions is mostly unknown. Here, we analyze the impact of spatially inhomogeneous electric fields on the membrane potential, the induced membrane field, and the induced current source density of one-dimensional cables as well as morphologically realistic neurons and discuss how the features of the extracellular field affect these quantities. We show through simulations that endogenous fields, associated with hippocampal theta and sharp waves, can greatly affect spike timing. These findings imply that local electric fields, generated by the cooperative action of brain cells, can influence the timing of neural activity.

Low intensity electric fields have been suggested to affect the ongoing neuronal activity in vitro and in human studies. However, the physiological mechanism of how weak electrical fields affect and interact with intact brain activity is not well understood. We performed in vivo extracellular and intracellular recordings from the neocortex and hippocampus of anesthetized rats and extracellular recordings in behaving rats. Electric fields were generated by sinusoid patterns at slow frequency (0.8, 1.25 or 1.7 Hz) via electrodes placed on the surface of the skull or the dura. Transcranial electric stimulation (TES) reliably entrained neurons in widespread cortical areas, including the hippocampus. The percentage of TES phase-locked neurons increased with stimulus intensity and depended on the behavioral state of the animal. TES-induced voltage gradient, as low as 1 mV/mm at the recording sites, was sufficient to phase-bias neuronal spiking. Intracellular recordings showed that both spiking and subthreshold activity were under the combined influence of TES forced fields and network activity. We suggest that TES in chronic preparations may be used for experimental and therapeutic control of brain activity.

Recent research has slowly corroded a belief that selective attention and consciousness are so tightly entangled that they cannot be individually examined. In this review, we summarize psychophysical and neurophysiological evidence for a dissociation between top-down attention and consciousness. The evidence includes recent findings that show subjects can attend to perceptually invisible objects. More contentious is the finding that subjects can become conscious of an isolated object, or the gist of the scene in the near absence of top-down attention; we critically re-examine the possibility of "complete" absence of top-down attention. We also cover the recent flurry of studies that utilized independent manipulation of attention and consciousness. These studies have shown paradoxical effects of attention, including examples where top-down attention and consciousness have opposing effects, leading us to strengthen and revise our previous views. Neuroimaging studies with EEG, MEG, and fMRI are uncovering the distinct neuronal correlates of selective attention and consciousness in dissociative paradigms. These findings point to a functional dissociation: attention as analyzer and consciousness as synthesizer. Separating the effects of selective visual attention from those of visual consciousness is of paramount importance to untangle the neural substrates of consciousness from those for attention.

Previous studies of eye gaze have shown that when looking at images containing human faces, observers tend to rapidly focus on the facial regions. But is this true of other high-level image features as well? We here investigate the extent to which natural scenes containing faces, text elements, and cell phones-as a suitable control-attract attention by tracking the eye movements of subjects in two types of tasks-free viewing and search. We observed that subjects in free-viewing conditions look at faces and text 16.6 and 11.1 times more than similar regions normalized for size and position of the face and text. In terms of attracting gaze, text is almost as effective as faces. Furthermore, it is difficult to avoid looking at faces and text even when doing so imposes a cost. We also found that subjects took longer in making their initial saccade when they were told to avoid faces/text and their saccades landed on a non-face/non-text object. We refine a well-known bottom-up computer model of saliency-driven attention that includes conspicuity maps for color, orientation, and intensity by adding high-level semantic information (i.e., the location of faces or text) and demonstrate that this significantly improves the ability to predict eye fixations in natural images. Our enhanced model's predictions yield an area under the ROC curve over 84% for images that contain faces or text when compared against the actual fixation pattern of subjects. This suggests that the primate visual system allocates attention using such an enhanced saliency map.

We demonstrate that human electrophysiological recordings of the local field potential (LFP) from intracranial electrodes, acquired from a variety of cerebral regions, show a ubiquitous 1/f(2) scaling within the power spectrum. We develop a quantitative model that treats the generation of these fields in an analogous way to that of electronic shot noise, and use this model to specifically address the cause of this 1/f(2) Brownian noise. The model gives way to two analytically tractable solutions, both displaying Brownian noise: 1) uncorrelated cells that display sharp initial activity, whose extracellular fields slowly decay in time and 2) rapidly firing, temporally correlated cells that generate UP-DOWN states.

During sustained viewing of an ambiguous stimulus, an individual's perceptual experience will generally switch between the different possible alternatives rather than stay fixed on one interpretation (perceptual rivalry). Here, we measured pupil diameter while subjects viewed different ambiguous visual and auditory stimuli. For all stimuli tested, pupil diameter increased just before the reported perceptual switch and the relative amount of dilation before this switch was a significant predictor of the subsequent duration of perceptual stability. These results could not be explained by blink or eye-movement effects, the motor response or stimulus driven changes in retinal input. Because pupil dilation reflects levels of norepinephrine (NE) released from the locus coeruleus (LC), we interpret these results as suggestive that the LC-NE complex may play the same role in perceptual selection as in behavioral decision making.

The most widely used method for detecting genome-wide protein-DNA interactions is chromatin immunoprecipitation on tiling microarrays, commonly known as ChIP-chip. Here, we conducted the first objective analysis of tiling array platforms, amplification procedures, and signal detection algorithms in a simulated ChIP-chip experiment. Mixtures of human genomic DNA and "spike-ins" comprised of nearly 100 human sequences at various concentrations were hybridized to four tiling array platforms by eight independent groups. Blind to the number of spike-ins, their locations, and the range of concentrations, each group made predictions of the spike-in locations. We found that microarray platform choice is not the primary determinant of overall performance. In fact, variation in performance between labs, protocols, and algorithms within the same array platform was greater than the variation in performance between array platforms. However, each array platform had unique performance characteristics that varied with tiling resolution and the number of replicates, which have implications for cost versus detection power. Long oligonucleotide arrays were slightly more sensitive at detecting very low enrichment. On all platforms, simple sequence repeats and genome redundancy tended to result in false positives. LM-PCR and WGA, the most popular sample amplification techniques, reproduced relative enrichment levels with high fidelity. Performance among signal detection algorithms was heavily dependent on array platform. The spike-in DNA samples and the data presented here provide a stable benchmark against which future ChIP platforms, protocol improvements, and analysis methods can be evaluated.

Retrograde amnesia following disruptions of hippocampal function is often temporally graded, with recent memories being more impaired. Evidence supports the existence of one or more neocortical long-term memory storage/retrieval site(s). Neurotoxic lesions of the medial prefrontal cortex (mPFC) or the dorsal hippocampus (DH) were made 1 day or 200 days following trace fear conditioning. Recently encoded trace fear memories were most disrupted by DH lesions, while remotely encoded trace and contextual memories were most disrupted by mPFC lesions. These data strongly support the consolidation theory of hippocampus function and implicate the mPFC as a site of long-term memory storage/retrieval.

Would you sell your soul on eBay? Right now, of course, you can't. But in some quarters it is taken for granted that within a generation, human beings-including you, if you can hang on for another 30 years or so-will have an alternative to death: being a ghost in a machine. You'll be able to upload your mind-your thoughts, memories, and personality-to a computer. And once you've reduced your consciousness to patterns of electrons, others will be able to copy it, edit it, sell it, or pirate it. It might be bundled with other electronic minds. And, of course, it could be deleted.

In natural vision both stimulus features and task-demands affect an observer's attention. However, the relationship between sensory-driven ("bottom-up") and task-dependent ("top-down") factors remains controversial: Can task-demands counteract strong sensory signals fully, quickly, and irrespective of bottom-up features? To measure attention under naturalistic conditions, we recorded eye-movements in human observers, while they viewed photographs of outdoor scenes. In the first experiment, smooth modulations of contrast biased the stimuli's sensory-driven saliency towards one side. In free-viewing, observers' eye-positions were immediately biased toward the high-contrast, i.e., high-saliency, side. However, this sensory-driven bias disappeared entirely when observers searched for a bull's-eye target embedded with equal probability to either side of the stimulus. When the target always occurred in the low-contrast side, observers' eye-positions were immediately biased towards this low-saliency side, i.e., the sensory-driven bias reversed. Hence, task-demands do not only override sensory-driven saliency but also actively countermand it. In a second experiment, a 5-Hz flicker replaced the contrast gradient. Whereas the bias was less persistent in free viewing, the overriding and reversal took longer to deploy. Hence, insufficient sensory-driven saliency cannot account for the bias reversal. In a third experiment, subjects searched for a spot of locally increased contrast ("oddity") instead of the bull's-eye ("template"). In contrast to the other conditions, a slight sensory-driven free-viewing bias prevails in this condition. In a fourth experiment, we demonstrate that at known locations template targets are detected faster than oddity targets, suggesting that the former induce a stronger top-down drive when used as search targets. Taken together, task-demands can override sensory-driven saliency in complex visual stimuli almost immediately, and the extent of overriding depends on the search target and the overridden feature, but not on the latter's free-viewing saliency.

Selective attention to a target yields faster and more accurate responses. Faster response times, in turn, are usually associated with increased subjective confidence. Could the decrease in reaction time in the presence of attention therefore simply reflect a shift toward more confident responses? We here addressed the extent to which attention modulates accuracy, processing speed, and confidence independently. To probe the effect of spatial attention on performance, we used two attentional manipulations of a visual orientation discrimination task. We demonstrate that spatial attention significantly increases accuracy, whereas subjective confidence measures reveal overconfidence in non-attended stimuli. At constant confidence levels, reaction times showed a significant decrease (by 15-49%, corresponding to 100-250 ms). This dissociation of objective performance and subjective confidence suggests that attention and awareness, as measured by confidence, are distinct, albeit related, phenomena.

This review examines recent advances in the study of brain correlates of consciousness. First, we briefly discuss some useful distinctions between consciousness and other brain functions. We then examine what has been learned by studying global changes in the level of consciousness, such as sleep, anesthesia, and seizures. Next we consider some of the most common paradigms used to study the neural correlates for specific conscious percepts and examine what recent findings say about the role of different brain regions in giving rise to consciousness for that percept. Then we discuss dynamic aspects of neural activity, such as sustained versus phasic activity, feedforward versus reentrant activity, and the role of neural synchronization. Finally, we briefly consider how a theoretical analysis of the fundamental properties of consciousness can usefully complement neurobiological studies.

Using functional magnetic resonance imaging, we found that cortical visual motion area MT+/V5 responded to auditory motion in two rare subjects who had been blind since early childhood and whose vision was partially recovered in adulthood. Visually normal control subjects did not show similar auditory responses. These auditory responses in MT+ were specific to motion compared with other complex auditory stimuli including frequency sweeps and speech. Thus, MT+ developed motion-specific responses to nonvisual input, suggesting that cross-modal plasticity can be influenced by the normal functional specialization of a cortical region. Regarding sight recovery after early blindness, our results further demonstrate that cross-modal responses coexist with regained visual responses within the visual cortex.

Neurons in the temporal lobe of both monkeys and humans show selective responses to classes of visual stimuli and even to specific individuals. In this study, we investigate the latency and selectivity of visually responsive neurons recorded from microelectrodes in the parahippocampal cortex, entorhinal cortex, hippocampus, and amygdala of human subjects during a visual object presentation task. During 96 experimental sessions in 35 subjects, we recorded from a total of 3278 neurons. Of these units, 398 responded selectively to one or more of the presented stimuli. Mean response latencies were substantially larger than those reported in monkeys. We observed a highly significant correlation between the latency and the selectivity of these neurons: the longer the latency the greater the selectivity. Particularly, parahippocampal neurons were found to respond significantly earlier and less selectively than those in the other three regions. Regional analysis showed significant correlations between latency and selectivity within the parahippocampal cortex, entorhinal cortex, and hippocampus, but not within the amygdala. The later and more selective responses tended to be generated by cells with sparse baseline firing rates and vice versa. Our results provide direct evidence for hierarchical processing of sensory information at the interface between the visual pathway and the limbic system, by which increasingly refined and specific representations of stimulus identity are generated over time along the anatomic pathways of the medial temporal lobe.

Butyrate is a histone deacetylase inhibitor (HDACi) with anti-neoplastic properties, which theoretically reactivates epigenetically silenced genes by increasing global histone acetylation. However, recent studies indicate that a similar number or even more genes are down-regulated than up-regulated by this drug. We treated hepatocarcinoma HepG2 cells with butyrate and characterized the levels of acetylation at DNA-bound histones H3 and H4 by ChIP-chip along the ENCODE regions. In contrast to the global increases of histone acetylation, many genomic regions close to transcription start sites were deacetylated after butyrate exposure. In order to validate these findings, we found that both butyrate and trichostatin A treatment resulted in histone deacetylation at selected regions, while nucleosome loss or changes in histone H3 lysine 4 trimethylation (H3K4me3) did not occur in such locations. Furthermore, similar histone deacetylation events were observed when colon adenocarcinoma HT-29 cells were treated with butyrate. In addition, genes with deacetylated promoters were down-regulated by butyrate, and this was mediated at the transcriptional level by affecting RNA polymerase II (POLR2A) initiation/elongation. Finally, the global increase in acetylated histones was preferentially localized to the nuclear periphery, indicating that it might not be associated to euchromatin. Our results are significant for the evaluation of HDACi as anti-tumourogenic drugs, suggesting that previous models of action might need to be revised, and provides an explanation for the frequently observed repression of many genes during HDACi treatment.

We generated high-resolution maps of histone H3 lysine 9/14 acetylation (H3ac), histone H4 lysine 5/8/12/16 acetylation (H4ac), and histone H3 at lysine 4 mono-, di-, and trimethylation (H3K4me1, H3K4me2, H3K4me3, respectively) across the ENCODE regions. Studying each modification in five human cell lines including the ENCODE Consortium common cell lines GM06990 (lymphoblastoid) and HeLa-S3, as well as K562, HFL-1, and MOLT4, we identified clear patterns of histone modification profiles with respect to genomic features. H3K4me3, H3K4me2, and H3ac modifications are tightly associated with the transcriptional start sites (TSSs) of genes, while H3K4me1 and H4ac have more widespread distributions. TSSs reveal characteristic patterns of both types of modification present and the position relative to TSSs. These patterns differ between active and inactive genes and in particular the state of H3K4me3 and H3ac modifications is highly predictive of gene activity. Away from TSSs, modification sites are enriched in H3K4me1 and relatively depleted in H3K4me3 and H3ac. Comparison between cell lines identified differences in the histone modification profiles associated with transcriptional differences between the cell lines. These results provide an overview of the functional relationship among histone modifications and gene expression in human cells.

We present a decomposition-based approach to managing incomplete information. We introduce world-set decompositions (WSDs), a space-efficient and complete representation system for finite sets of worlds. We study the problem of efficiently evaluating relational algebra queries on world-sets represented by WSDs. We also evaluate our technique experimentally in a large census data scenario and show that it is both scalable and efficient.

We investigated the representation of visual inputs by multiple simultaneously recorded single neurons in the human medial temporal lobe, using their firing rates to infer which images were shown to subjects. The selectivity of these neurons was quantified with a novel measure. About four spikes per neuron, triggered between 300 and 600 ms after image onset in a handful of units (7.8 on average), predicted the identity of images far above chance. Decoding performance increased linearly with the number of units considered, peaked between 400 and 500 ms, did not improve when considering correlations among simultaneously recorded units, and generalized to very different images. The feasibility of decoding sensory information from human extracellular recordings has implications for the development of brain-machine interfaces.

What do we see when we glance at a natural scene and how does it change as the glance becomes longer? We asked naive subjects to report in a free-form format what they saw when looking at briefly presented real-life photographs. Our subjects received no specific information as to the content of each stimulus. Thus, our paradigm differs from previous studies where subjects were cued before a picture was presented and/or were probed with multiple-choice questions. In the first stage, 90 novel grayscale photographs were foveally shown to a group of 22 native-English-speaking subjects. The presentation time was chosen at random from a set of seven possible times (from 27 to 500 ms). A perceptual mask followed each photograph immediately. After each presentation, subjects reported what they had just seen as completely and truthfully as possible. In the second stage, another group of naive individuals was instructed to score each of the descriptions produced by the subjects in the first stage. Individual scores were assigned to more than a hundred different attributes. We show that within a single glance, much object- and scene-level information is perceived by human subjects. The richness of our perception, though, seems asymmetrical. Subjects tend to have a propensity toward perceiving natural scenes as being outdoor rather than indoor. The reporting of sensory- or feature-level information of a scene (such as shading and shape) consistently precedes the reporting of the semantic-level information. But once subjects recognize more semantic-level components of a scene, there is little evidence suggesting any bias toward either scene-level or object-level recognition.

Consciousness is a puzzling, state-dependent property of certain types of complex, biological, adaptive, and highly interconnected systems. A science of consciousness must strive to explain the exact relationship between phenomenal, mental states and brain states. This is the heart of the classical mind-body problem: What is the nature of the relationship between the immaterial, conscious mind and its physical basis in the electrochemical interactions in the body? Brain scientists are exploiting a number of empirical approaches to shed light on the neural basis of consciousness. This article reviews these approaches and summarizes what has been learnt.

A new bottom-up visual saliency model, Graph-Based Visual Saliency (GBVS), is proposed. It consists of two steps: first forming activation maps on certain feature&#13;\nchannels, and then normalizing them in a way which highlights conspicuity and admits combination with other maps. The model is simple, and biologically plausible&#13;\ninsofar as it is naturally parallelized. This model powerfully predicts human fixations on 749 variations of 108 natural images, achieving 98% of the ROC area&#13;\nof a human-based control, whereas the classical algorithms of Itti &amp; Koch ([2], [3], [4]) achieve only 84%.

Under natural viewing conditions, human observers shift their gaze to allocate processing resources to subsets of the visual input. Many computational models try to predict such voluntary eye and attentional shifts. Although the important role of high level stimulus properties (e.g., semantic information) in search stands undisputed, most models are based on low-level image properties. We here demonstrate that a combined model of face detection and low-level saliency significantly outperforms a low-level model in predicting locations humans fixate on, based on eye-movement recordings of humans observing photographs of natural scenes, most of which contained at least one person. Observers, even when not instructed to look for anything particular, fixate on a face with a probability of over 80% within their first two fixations; furthermore, they exhibit more similar scanpaths when faces are present. Remarkably, our model’s predictive performance in images that do not contain faces is not impaired, and is even improved in some cases by spurious face detector responses.

Although extracellular unit recording is typically used for the detection of spike occurrences, it also has the theoretical ability to report about what are typically considered intracellular features of the action potential. We address this theoretical ability by developing a model system that captures features of experimentally recorded simultaneous intracellular and extracellular recordings of CA1 pyramidal neurons. We use the line source approximation method of Holt and Koch to model the extracellular action potential (EAP) voltage resulting from the spiking activity of individual neurons. We compare the simultaneous intracellular and extracellular recordings of CA1 pyramidal neurons recorded in vivo with model predictions for the same cells reconstructed and simulated with compartmental models. The model accurately reproduces both the waveform and the amplitude of the EAPs, although it was difficult to achieve simultaneous good matches on both the intracellular and extracellular waveforms. This suggests that accounting for the EAP waveform provides a considerable constraint on the overall model. The developed model explains how and why the waveform varies with electrode position relative to the recorded cell. Interestingly, each cell's dendritic morphology had very little impact on the EAP waveform. The model also demonstrates that the varied composition of ionic currents in different cells is reflected in the features of the EAP.

When conflicting images are presented to the corresponding regions of the two eyes, only one image may be consciously perceived. In binocular rivalry (BR), two images alternate in phenomenal visibility; even a salient image is eventually suppressed by an image of low saliency. Recently, N. Tsuchiya and C. Koch (2005) reported a technique called continuous flash suppression (CFS), extending the suppression duration more than 10-fold. Here, we investigated the depth of this prolonged form of interocular suppression as well as conventional BR and flash suppression (FS) using a probe detection task. Compared to monocular viewing condition, CFS elevated detection thresholds more than 20-fold, whereas BR did so by 3-fold. In subsequent experiments, we dissected CFS into several components. By manipulating the number and timing of flashes with respect to the probe, we found that the stronger suppression in CFS is not due to summation between BR and FS but is caused by the summation of the suppression due to multiple flashes. Our results support the view that CFS is not a stronger version of BR but is due to the accumulated suppressive effects of multiple flashes.

Recent experiments characterized individual neurons in the human medial temporal lobe with remarkably selective, invariant, and explicit responses to images of famous individuals or landmark buildings. Here, we used a probabilistic analysis to show that these data are consistent with a sparse code in which neurons respond in a selective manner to a small fraction of stimuli.

Continuously moving objects sometimes appear to spontaneously reverse their motion direction. The mechanisms underlying this bistable phenomenon (the "continuous wagon wheel illusion") are heavily debated, but one interpretation suggests that motion information is perceived in discrete episodes at a rate between 10 and 15 Hz. Here, we asked observers to report the perceived direction of a continuously rotating wheel while 32-channel electroencephalogram (EEG) was recorded. We then separated periods of perceived true from illusory (reversed) motion and compared the EEG power spectrum under these two perceptually distinct yet physically identical conditions. The only reliable difference was observed approximately 13 Hz over centroparietal electrodes, independent of the temporal frequency of the wheel. Thus, it is likely to reflect internal processes rather than purely stimulus-driven activity. EEG power (13 Hz) decreased before the onset of illusory motion and increased before transitions back to real motion. Using this relationship, it was possible to predict above chance, on a trial-by-trial basis, the direction of the upcoming perceptual transition. These data are compatible with the idea that motion perception occurs in snapshots <100 ms in duration.

In movies or on TV, a wheel can seem to rotate backwards, due to the temporal subsampling inherent in the recording process (the wagon wheel illusion). Surprisingly, this effect has also been reported under continuous light, suggesting that our visual system, too, might sample motion in discrete "snapshots." Recently, these results and their interpretation have been challenged. Here, we investigate the continuous wagon wheel illusion as a form of bistable percept. We observe a strong temporal frequency dependence: the illusion is maximal at alternation rates around 10 Hz but shows no spatial frequency dependence. We introduce an objective method, based on unbalanced counterphase gratings, for measuring this phenomenon and demonstrate that the effect critically depends on attention: the continuous wagon wheel illusion was almost abolished in the absence of focused attention. A motion-energy model, coupled with attention-dependent temporal subsampling of the perceptual stream at rates between 10 and 20 Hz, can quantitatively account for the observed data.

It was recently demonstrated that detecting target objects (e.g., animals) in natural scenes can be done in a dual-task paradigm, in the near absence of spatial attention. Under the same conditions, subjects were unable to perform apparently simpler tasks involving synthetic stimuli (e.g., discriminating randomly rotated Ts and Ls, or a bisected colour disc and its mirror image). Classical theories predict that attention is more critical for the recognition of complex stimuli that cannot be easily separated on a single feature dimension. Therefore, these puzzling results have raised a number of questions. If it is not the complexity of a stimulus, what then determines the recognition task's attentional requirements? How does this differ between natural and artificial stimuli? What can these observations tell us about the mechanism of natural scene processing as well as its relation to attention? Here we show that removing colour information, or doubling the amount of information to be analysed, failed to make the natural scene categorization tasks significantly more “attention demanding”. Conversely, increasing discriminability or predictability did not diminish the need for attention in the case of synthetic stimuli. However, when the familiar letters, such as Ts or Ls, were presented upright, full attention was no longer required for discrimination. This suggests that familiarity and meaningfulness might be among the factors that determine attentional requirements for both natural and synthetic stimuli.

The claustrum is a thin, irregular, sheet-like neuronal structure hidden beneath the inner surface of the neocortex in the general region of the insula. Its function is enigmatic. Its anatomy is quite remarkable in that it receives input from almost all regions of cortex and projects back to almost all regions of cortex. We here briefly summarize what is known about the claustrum, speculate on its possible relationship to the processes that give rise to integrated conscious percepts, propose mechanisms that enable information to travel widely within the claustrum and discuss experiments to address these questions.

Neurones are noisy elements. Noise arises from both intrinsic and extrinsic sources, and manifests itself as fluctuations in the membrane potential. These fluctuations limit the accuracy of a neurone's output but have also been suggested to play a computational role. We present a detailed study of the amplitude and spectrum of voltage noise recorded at the soma of layer IV-V pyramidal neurones in slices taken from rat neocortex. The dependence of the noise on holding potential, synaptic activity and Na+ conductance is systematically analysed. We demonstrate that voltage noise increases non-linearly as the cell depolarizes (from a standard deviation (s.d.) of 0.19 mV at -75 mV to an s.d. of 0.54 mV at -55 mV). The increase in voltage noise is accompanied by an increase in the cell impedance, due to voltage dependence of Na+ conductance. The impedance increase accounts for the majority (70%) of the voltage noise increase. The increase in voltage noise and impedance is restricted to the low-frequency range (0.2-2 Hz). At the high frequency range (5-100 Hz) the voltage noise is dominated by synaptic activity. In our slice preparation, synaptic noise has little effect on the cell impedance. A minimal model reproduces qualitatively these data. Our results imply that ion channel noise contributes significantly to membrane voltage fluctuations at the subthreshold voltage range, and that Na+ conductance plays a key role in determining the amplitude of this noise by acting as a voltage-dependent amplifier of low-frequency transients.

For oceanographic research, remotely operated underwater vehicles (ROVs) routinely record several hours of video material each day. Manual processing of such large amounts of video has become a major bottleneck for scientific research based on this data. We have developed an automated system that detects and tracks objects that are of potential interest for human video annotators. By pre-selecting salient targets for track initiation using a selective attention algorithm, we reduce the complexity of multi-target tracking, in particular of the assignment problem. Detection of low-contrast translucent targets is difficult due to variable lighting conditions and the presence of ubiquitous noise from high-contrast organic debris ("marine snow") particles. We describe the methods we developed to overcome these issues and report our results of processing ROV video data.

A key problem in learning multiple objects from unlabeled images is that it is a priori impossible to tell which part of the image corresponds to each individual object, and which part is irrelevant clutter which is not associated to the objects. We investigate empirically to what extent pure bottom-up attention can extract useful information about the location, size and shape of objects from images and demonstrate how this information can be utilized to enable unsupervised learning of objects from unlabeled images. Our experiments demonstrate that the proposed approach to using bottom-up attention is indeed useful for a variety of applications.

The attentional cost associated with the visual discrimination of the gender of a face was investigated. Participants performed a face-gender discrimination task either alone (single-task) or concurrently (dual-task) with a known attentional demanding task (5-letter T/L discrimination). Overall performance on face-gender discrimination suffered remarkably little under the dual-task condition compared to the single-task condition. Similar results were obtained in experiments that controlled for potential training effects or the use of low-level cues in this discrimination task. Our results provide further evidence against the notion that only low-level representations can be accessed outside the focus of attention.

In this chapter from his book, a Caltech neurobiologist talks about computer consciousness, zombies, and other things.

Previous studies of associative learning implicate higher-level cognitive processes in some forms of classical conditioning. An ongoing debate is concerned with the extent to which attention and awareness are necessary for trace but not delay eye-blink conditioning [Clark, R. E. & Squire, L. R. (1998) Science 280, 77-81; Lovibond, P. F. & Shanks, D. (2002) J. Exp. Psychol. Anim. Behav. Processes 28, 38-42]. In trace conditioning, a short interval is interposed between the termination of the conditioned stimulus (CS) and the onset of the unconditioned stimulus (US). In delay conditioning, the CS and US overlap. We here investigate the extent to which human classical fear conditioning depends on working memory. Subjects had to carry out an n-back task, requiring tracking an item 1 or 2 back in a sequentially presented list of numbers, while simultaneously being tested for their ability to associate auditory cues with shocks under a variety of conditions (single-cue versus differential; delay versus trace; no task versus 0-, 1-, and 2-back). Differential delay conditioning proved to be more resilient than differential trace conditioning but does show a reduction due to task interference similar in slope to that found in trace conditioning. Explicit knowledge of the stimulus contingency facilitates but does not guarantee trace conditioning. Only the single-cue delay protocol shows conditioning during the more difficult working memory task. Our findings suggest that the larger the cognitive demands on the system, the less likely conditioning occurs. A postexperimental questionnaire showed a positive correlation between conditioning and awareness for differential trace conditioning extinction.

Higher cognitive functions such as attention have been difficult to model in genetically tractable organisms. In humans, attention-distracting stimuli interfere with trace but not delay conditioning, two forms of associative learning. Attention has also been correlated with activation of anterior cingulate cortex (ACC), but its functional significance is unclear. Here we show that a visual distractor interferes selectively with trace but not delay auditory fear conditioning in mice. Trace conditioning is associated with increased neuronal activity in ACC, as assayed by relative levels of c-fos expression, and is selectively impaired by lesions of this structure. The effects of the ACC lesions are unlikely to be caused by indirect impairment of the hippocampus, which is required for mnemonic aspects of trace conditioning. These data suggest that trace conditioning may be useful for studying neural substrates of attention in mice, and implicate the ACC as one such substrate.

The ventral visual pathway implements object recognition and categorization in a hierarchy of processing areas with neuronal selectivities of increasing complexity. The presence of massive feedback connections within this hierarchy raises the possibility that normal visual processing relies on the use of computational loops. It is not known, however, whether object recognition can be performed at all without such loops (i.e., in a purely feed-forward mode). By analyzing the time course of reaction times in a masked natural scene categorization paradigm, we show that the human visual system can generate selective motor responses based on a single feed-forward pass. We confirm these results using a more constrained letter discrimination task, in which the rapid succession of a target and mask is actually perceived as a distractor. We show that a masked stimulus presented for only 26 msec - and often not consciously perceived - can fully determine the earliest selective motor responses: The neural representations of the stimulus and mask are thus kept separated during a short period corresponding to the feed-forward "sweep." Therefore, feedback loops do not appear to be "mandatory" for visual processing. Rather, we found that such loops allow the masked stimulus to reverberate in the visual system and affect behavior for nearly 150 msec after the feed-forward sweep.

Visual information from the environment is transformed into perceptual sensations through several stages of neuronal processing. Flash suppression constitutes a striking example in which the same retinal input can give rise to two different conscious visual percepts. We directly recorded the responses of individual neurons during flash suppression in the human amygdala, entorhinal cortex, hippocampus, and parahippocampal gyrus, allowing us to explore the neuronal responses in untrained subjects at a high spatial and temporal resolution in the medial temporal lobe. Subjects were patients with pharmacologically intractable epilepsy implanted with depth electrodes to localize the seizure onset focus. We observed that the activity of two thirds of all visually selective neurons followed the perceptual alternations rather than the retinal input. None of the selective neurons responded to a perceptually suppressed stimulus. Therefore, the activity of most individual neurons in the medial temporal lobe of naive human subjects directly correlates with the phenomenal visual experience.

What can we see when we do not pay attention? It is well known that we can be "blind" even to major aspects of natural scenes when we attend elsewhere. The only tasks that do not need attention appear to be carried out in the early stages of the visual system. Contrary to this common belief, we report that subjects can rapidly detect animals or vehicles in briefly presented novel natural scenes while simultaneously performing another attentionally demanding task. By comparison, they are unable to discriminate large T's from L's, or bisected two-color disks from their mirror images under the same conditions. We conclude that some visual tasks associated with "high-level" cortical areas may proceed in the near absence of attention.

We characterize a class of spatio-temporal illusions with two complementary properties. Firstly, if a vernier stimulus is flashed for a short time on a monitor and is followed immediately by a grating, the latter can express features of the vernier, such as its offset, its orientation, or its motion (feature inheritance). Yet the vernier stimulus itself remains perceptually invisible. Secondly, the vernier can be rendered visible by presenting gratings with a larger number of elements (shine-through). Under these conditions, subjects perceive two independent "objects" each carrying their own features. Transition between these two domains can be effected by subtle changes in the spatio-temporal layout of the grating. This should allow psychophysicists and electrophysiologists to investigate feature binding in a precise and quantitative manner.

Although visual attention is known to modulate brain activity in the posterior parietal, prefrontal, and visual sensory areas, the unique roles of these areas in the control of attentional resources have remained unclear. Here, we report a dissociation in the response profiles of these areas. In a parametric functional magnetic resonance imaging (fMRI) study, subjects performed a covert motion-tracking task, in which we manipulated "attentional load" by varying the number of tracked balls. While strong effects of attention--independent of attentional load--were widespread, robust linear increases of brain activity with number of balls tracked were seen primarily in the posterior parietal areas, including the intraparietal sulcus (IPS) and superior parietal lobule (SPL). Thus, variations in attentional load revealed different response profiles in sensory areas as compared to control areas. Our results suggest a general role for posterior parietal areas in the deployment of visual of attentional resources.

OBJECTIVES: To evaluate the neural correlates of attention and working memory deficits in patients with HIV-1. METHOD: fMRI was used to evaluate brain activity in 11 patients with HIV and 11 age-, sex-, education-, and handedness-matched seronegative subjects, while performing a battery of tasks that required different levels of attention for working memory. RESULTS: Patients with HIV showed greater brain activation (blood oxygenation level dependent signal changes) in some regions compared with control subjects while performing the same tasks. For the simpler tasks, patients with HIV showed greater activation in the parietal regions. However, with more difficult tasks, patients with HIV showed greater activation additionally in the frontal lobes. Reaction times during these tasks were slower but accuracy was similar in the patients with HIV compared with control subjects. CONCLUSION: Injury to the neural substrate caused by HIV infection may necessitate greater attentional modulation of the neural circuits, hence a greater use of the brain reserve; additional activation of the frontal lobes is required to perform the more complex tasks. The task-dependent increased frontal activation in patients with HIV suggests that the neural correlate of attentional deficits may be excessive attentional modulation as a result of frontostriatal brain injury.

Gender differences in brain activation during working memory tasks were examined with fMRI. Seventeen right-handed subjects (nine males, eight females) were studied with four different verbal working memory tasks of varying difficulty using whole brain echo-planar fMRI. Consistent with prior studies, we observed activation of the lateral prefrontal cortices (LPFC), the parietal cortices (PC), and additionally, caudate activation in both sexes. The volume of activated brain tissue increased with increasing task difficulty. For all four tasks, the male subjects showed bilateral activation or right-sided dominance (LPFC, PC and caudate), whereas females showed activation predominantly in the left hemisphere. The task performance data demonstrated higher accuracy and slightly slower reaction times for the female subjects. Our results show a highly significant (p < 0.001) gender differences in the functional organization of the brain for working memory. These gender-specific differences in functional organization of the brain may be due to gender-differences in problem solving strategies or the neurodevelopment. Therefore, gender matching or stratification is required for studies of brain function using imaging techniques.

When we observe our visual environment, we do not perceive all its components as being equally interesting. Some objects automatically and effortlessly “pop-out” from their surroundings, that is, they draw our visual attention, in a “ bottom-up” manner, towards them. In a first approximation, focal visual attention acts as a rapidly shiftable “spotlight,” which allows only the selected information to reach higher levels of processing and representation. Most models of the bottom-up control of attention are based on the concept of a saliency map, that is, an explicit two-dimensional map that encodes the conspicuity of objects in the visual environment. Competition among neurons in this map gives rise to a single winning location that corresponds to the next attended target. Inhibiting this location automatically allows the system to attend to the next most salient location. A first body of work in this thesis describes a detailed computer implementation of such a scheme, focusing on the problem of combining information across modalities, here orientation, intensity and color information, in a purely stimulus-driven manner. The model is applied to common psychophysical stimuli as well as to very demanding visual search tasks. Its successful performance is used to address the extent to which the primate visual system carries out visual search via one or more such saliency maps and how this can be tested. We next address the question of what happens once our attention is focused onto a restricted part of our visual field. There is mounting experimental evidence that attention is far more sophisticated than a simple feed-forward spatially-selective filtering process. Indeed, visual processing appears to be significantly different inside the attentional spotlight than outside. That is, in addition to its properties as a feed-forward information processing and transmission bottleneck, focal visual attention feeds back and locally modulates, in a “top-down” manner, the visual processing and representation of selected objects. The second body of work presented in this thesis is concerned with a detailed computational model of basic pattern vision in humans and its modulation by top-down attention. We start by acquiring a complete dataset of five different simple psychophysical experiments, including discriminations of contrast, orientation and spatial frequency of simple pattern stimuli by human observers. This experimental dataset places strict constraints on our model of early pattern vision. The model, however, is eventually able to reproduce the entire dataset while assuming plausible neurobiological components. The model is further applied to existing psychophysical data which demonstrates how top-down attention alters performance in these simple psychophysical discrimination experiments. Our model is able to quantitatively account for all observations by assuming that attention strengthens the non-linear cortical interactions among visual neurons.

Bottom-up or saliency-based visual attention allows primates to detect non-specific conspicuous targets in cluttered scenes. A classical metaphor, derived from electrophysiological and psychophysical studies, describes attention as a rapidly shiftable 'spotlight'. The model described here reproduces the attentional scanpaths of this spotlight: Simple multi-scale 'feature maps' detect local spatial discontinuities in intensity, color, orientation or optical flow, and are combined into a unique 'master' or 'saliency' map. the saliency map is sequentially scanned, in order of decreasing saliency, by the focus of attention. We study the problem of combining feature maps, from different visual modalities and with unrelated dynamic ranges, into a unique saliency map. Four combination strategies are compared using three databases of natural color images: (1) Simple normalized summation, (2) linear combination with learned weights, (3) global non-linear normalization followed by summation, and (4) local non-linear competition between salient locations. Performance was measured as the number of false detections before the most salient target was found. Strategy (1) always yielded poorest performance and (2) best performance, with a 3- to 8-fold improvement in time to find a salient target. However, (2) yielded specialized systems with poor generations. Interestingly, strategy (4) and its simplified, computationally efficient approximation (3) yielded significantly better performance than (1), with up to 4-fold improvement, while preserving generality.

Advances in the neurosciences have revealed the staggering complexity of even "simple" nervous systems. This is reflected in their function, their evolutionary history, their structure, and the coding schemes they use to represent information. These four viewpoints need all play a role in any future science of "brain complexity."

Encoding of visual information by LGN bursts. Thalamic relay cells respond to visual stimuli either in burst mode, as a result of activation of a low-threshold Ca2+ conductance, or in tonic mode, when this conductance is inactive. We investigated the role of these two response modes for the encoding of the time course of dynamic visual stimuli, based on extracellular recordings of 35 relay cells from the lateral geniculate nucleus of anesthetized cats. We presented a spatially optimized visual stimulus whose contrast fluctuated randomly in time with frequencies of up to 32 Hz. We estimated the visual information in the neural responses using a linear stimulus reconstruction method. Both burst and tonic spikes carried information about stimulus contrast, exceeding one bit per action potential for the highest variance stimuli. The "meaning" of an action potential, i.e., the optimal estimate of the stimulus at times preceding a spike, was similar for burst and tonic spikes. In within-trial comparisons, tonic spikes carried about twice as much information per action potential as bursts, but bursts as unitary events encoded about three times more information per event than tonic spikes. The coding efficiency of a neuron for a particular stimulus is defined as the fraction of the neural coding capacity that carries stimulus information. Based on a lower bound estimate of coding efficiency, bursts had approximately 1.5-fold higher efficiency than tonic spikes, or 3-fold if bursts were considered unitary events. Our main conclusion is that both bursts and tonic spikes encode stimulus information efficiently, which rules out the hypothesis that bursts are nonvisual responses.

In recent theoretical approaches addressing the problem of neural coding, tools from statistical estimation and information theory have been applied to quantify the ability of neurons to transmit information through their spike outputs. These techniques, though fairly general, ignore the specific nature of neuronal processing in terms of its known biophysical properties. However, a systematic study of processing at various stages in a biophysically faithful model of a single neuron can identify the role of each stage in information transfer. Toward this end, we carry out a theoretical analysis of the information loss of a synaptic signal propagating along a linear, one-dimensional, weakly active cable due to neuronal noise sources along the way, using both a signal reconstruction and a signal detection paradigm. Here we begin such an analysis by quantitatively characterizing three sources of membrane noise: (1) thermal noise due to the passive membrane resistance, (2) noise due to stochastic openings and closings of voltage-gated membrane channels (NA+ and K+), and (3) noise due to random, background synaptic activity. Using analytical expressions for the power spectral densities of these noise sources, we compare their magnitudes in the case of a patch of membrane from a cortical pyramidal cell and explore their dependence on different biophysical parameters.

Neural network research often builds on the fiction that neurons are simple linear threshold units, completely neglecting the highly dynamic and complex nature of synapses, dendrites, and voltage-dependent ionic currents. Biophysics of Computation: Information Processing in Single Neurons challenges this notion, using richly detailed experimental and theoretical findings from cellular biophysics to explain the repertoire of computational functions available to single neurons. The author shows how individual nerve cells can multiply, integrate, or delay synaptic inputs and how information can be encoded in the voltage across the membrane, in the intracellular calcium concentration, or in the timing of individual spikes. Key topics covered include the linear cable equation; cable theory as applied to passive dendritic trees and dendritic spines; chemical and electrical synapses and how to treat them from a computational point of view; nonlinear interactions of synaptic input in passive and active dendritic trees; the Hodgkin-Huxley model of action potential generation and propagation; phase space analysis; linking stochastic ionic channels to membrane-dependent currents; calcium and potassium currents and their role in information processing; the role of diffusion, buffering and binding of calcium, and other messenger systems in information processing and storage; short- and long-term models of synaptic plasticity; simplified models of single cells; stochastic aspects of neuronal firing; the nature of the neuronal code; and unconventional models of sub-cellular computation. Biophysics of Computation: Information Processing in Single Neurons serves as an ideal text for advanced undergraduate and graduate courses in cellular biophysics, computational neuroscience, and neural networks, and will appeal to students and professionals in neuroscience, electrical and computer engineering, and physics.

A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented. Multiscale image features are combined into a single topographical saliency map. A dynamical neural network then selects attended locations in order of decreasing saliency. The system breaks down the complex problem of scene understanding by rapidly selecting, in a computationally efficient manner, conspicuous locations to be analyzed in detail.

In most sensory systems, higher order central neurons extract those stimulus features from the sensory periphery that are behaviorally relevant (e.g.,Marr, 1982; Heiligenberg, 1991). Recent studies have quantified the time-varying information carried by spike trains of sensory neurons in various systems using stimulus estimation methods (Bialek et al., 1991; Wessel et al., 1996). Here, we address the question of how this information is transferred from the sensory neuron level to higher order neurons across multiple sensory maps by using the electrosensory system in weakly electric fish as a model. To determine how electric field amplitude modulations are temporally encoded and processed at two subsequent stages of the amplitude coding pathway, we recorded the responses of P-type afferents and E- and I-type pyramidal cells in the electrosensory lateral line lobe (ELL) to random distortions of a mimic of the fish's own electric field. Cells in two of the three somatotopically organized ELL maps were studied (centromedial and lateral) (Maler, 1979; Carr and Maler, 1986). Linear and second order nonlinear stimulus estimation methods indicated that in contrast to P-receptor afferents, pyramidal cells did not reliably encode time-varying information about any function of the stimulus obtained by linear filtering and half-wave rectification. Two pattern classifiers were applied to discriminate stimulus waveforms preceding the occurrence or nonoccurrence of pyramidal cell spikes in response to the stimulus. These signal-detection methods revealed that pyramidal cells reliably encoded the presence of upstrokes and downstrokes in random amplitude modulations by short bursts of spikes. Furthermore, among the different cell types in the ELL, I-type pyramidal cells in the centromedial map performed a better pattern-recognition task than those in the lateral map and than E-type pyramidal cells in either map.

From the Publisher: This book serves as a handbook of computational methods and techniques for modeling the functional properties of single and groups of nerve cells.

What is the relationship between the temporal jitter in the arrival times of individual synaptic inputs to a neuron and the resultant jitter in its output spike? We report that the rise time of firing rates of cells in striate and extrastriate visual cortex in the macaque monkey remain equally sharp at different stages of processing. Furthermore, as observed by others, multiunit recordings from single units in the primate frontal lobe reveal a strong peak in their cross-correlation in the 10-150 msec range with very small temporal jitter (on the order of 1 msec). We explain these results using numerical models to study the relationship between the temporal jitter in excitatory and inhibitory synaptic input and the variability in the spike output timing in integrate-and-fire units and in a biophysically and anatomically detailed model of a cortical pyramidal cell. We conclude that under physiological circumstances, the standard deviation in the output jitter is linearly related to the standard deviation in the input jitter, with a constant of less than one. Thus, the timing jitter in successive layers of such neurons will converge to a small value dictated by the jitter in axonal propagation times.

We present two algorithms for estimating the velocity of a visual stimulus and their implementations with analog circuits using CMOS VLSI technology. Both are instances of so-called token methods, where velocity is computed by identifying particular features in the image at different locations; in our algorithms, these features are abrupt temporal changes in image irradiance. Our circuits integrate photoreceptors and associated electronics for computing motion onto a single chip and unambiguously extract bidirectional velocity for stimuli of high and intermediate contrasts over considerable irradiance and velocity ranges. At low contrasts, the output signal for a given velocity tends to decrease gracefully with contrast, while direction-selectivity is maintained. The individual motion-sensing cells are compact and highly suitable for use in dense 1-D or 2-D imaging arrays.

Shunting inhibition, a conductance increase with a reversal potential close to the resting potential of the cell, has been shown to have a divisive effect on subthreshold excitatory postsynaptic potential amplitudes. It has therefore been assumed to have the same divisive effect on firing rates. We show that shunting inhibition actually has a subtractive effect on the firing rate in most circumstances. Averaged over several interspike intervals, the spiking mechanism effectively clamps the somatic membrane potential to a value significantly above the resting potential, so that the current through the shunting conductance is approximately independent of the firing rate. This leads to a subtractive rather than a divisive effect. In addition, at distal synapses, shunting inhibition will also have an approximately subtractive effect if the excitatory conductance is not small compared to the inhibitory conductance. Therefore regulating a cell's passive membrane conductance-for instance, via massive feedback-is not an adequate mechanism for normalizing or scaling its output.

That the cerebral cortex processes information at prodigious speeds cannot be doubted. Yet the passive time constant, tau(m), of neurons, often thought of as a measure of the neuron's "response time' to synaptic input, is relatively long. In the 1950s, tau(m) was estimated to be only a few milliseconds for mammalian central neurons; with improvement in recording techniques, its estimated value grew over the years and it now stands near 20-100 msec. However, as we will argue here, the functional meaning of tau(m) is ambiguous. On the basis of a newly introduced definition of local delay, we show that the time window for synaptic integration in passive dendritic trees can be much smaller than the time constant. We argue that the voltage response to very brief synaptic inputs is essentially independent of tau(m). We discuss how tau(m) can change dynamically with the global activity of the network, as well as the difficulties of defining a time constant in structures with voltage-dependent elements. We conclude that the classically defined tau(m) only provides a very rough estimate, typically an overestimate, of the response time of neurons and that alternative measures are required to capture the dependency of the time course of the membrane potential on ligand-gated and/or voltage-dependent membrane conductances.

1. In neocortical slices, the majority of neurons fire quite regularly in response to constant current injections. But neurons in the intact animal fire irregularly in response to constant current injection as well as to visual stimuli. 2. To quantify this observation, we developed a new measure of variability, which compares only adjacent interspike intervals and is therefore less sensitive to rate variations than existing measures such as the coefficient of variation of interspike intervals. 3. We find that the variability of firing is much higher in cells of primary visual cortex in the anesthetized cat than in slice. The response to current injected from an intracellular electrode in vivo is also variable, but slightly more regular and less bursty than in response to visual stimuli. 4. Using a new technique for analyzing the variability of integrate-and-fire neurons, we prove that this behavior is consistent with a simple integrate-and-fire model receiving a large amount of synaptic background activity, but not with a noisy spiking mechanism.

1. The coding of time-varying electric fields in the weakly electric fish, Eigenmannia, was investigated in a quantitative manner. The activity of single P-type electroreceptor afferents was recorded while the amplitude of an externally applied sinusoidal electric field was stochastically modulated. The amplitude modulation waveform (i.e., the stimulus) was reconstructed from the spike trains by mean square estimation. 2. From the stimulus and the reconstructions we calculated the following: 1) the signal-to-noise ratio and thus an effective temporal bandwidth of the units; 2) the coding fraction, i.e., a measure of the fraction of the time-varying stimulus encoded in single spike trains; and 3) the mutual information provided by the reconstructions about the stimulus. 3. Signal-to-noise ratios as high as 7:1 were observed and the bandwidth ranged from 0 up to 200 Hz, consistent with the limit imposed by the sampling theorem. Reducing the cutoff frequency of the stimulus increased the signal-to-noise ratio at low frequencies, indicating a nonlinearity in the receptors' response. 4. The coding fraction and the rate of mutual information transmission increased in parallel with the standard deviation (i.e., the contrast) of the stimulus as well as the mean firing rate of the units. Significant encoding occurred 20-40 Hz above the spontaneous discharge of a unit. 5. When the temporal cutoff frequency of the stimulus was increased between 80 and 400 Hz, 1) the coding fraction decreased, 2) the rate of mutual information transmission remained constant over the same frequency range, and 3) the reconstructed filter changed. This is in agreement with predictions obtained in a simplified neuronal model. 6. Our results suggest that 1) the information transmitted by single spike trains of primary electrosensory afferents to higherorder neurons in the fish brain depends on the contrast and the cutoff frequency of the stimulus as well as on the mean firing rate of the units; and 2) under optimal conditions, more than half of the information about a Gaussian stimulus that can in principle be encoded is carried in single spike trains of P-type afferents at rates up to 200 bits per second.

How reliably do action potentials in cortical neurons encode information about a visual stimulus? Most physiological studies do not weigh the occurrences of particular action potentials as significant but treat them only as reflections of average neuronal excitation. We report that single neurons recorded in a previous study by Newsome et al. (1989; see also Britten et al. 1992) from cortical area MT in the behaving monkey respond to dynamic and unpredictable motion stimuli with a markedly reproducible temporal modulation that is precise to a few milliseconds. This temporal modulation is stimulus dependent, being present for highly dynamic random motion but absent when the stimulus translates rigidly.

The majority of synapses in the mammalian cortex originate from cortical neurons. Indeed, the largest input to cortical cells comes from neighboring excitatory cells. However, most models of cortical development and processing do not reflect the anatomy and physiology of feedback excitation and are restricted to serial feedforward excitation. This report describes how populations of neurons in cat visual cortex can use excitatory feedback, characterized as an effective "network conductance", to amplify their feedforward input signals and demonstrates how neuronal discharge can be kept proportional to stimulus strength despite strong, recurrent connections that threaten to cause runaway excitation. These principles are incorporated into models of cortical direction and orientation selectivity that emphasize the basic design principles of cortical architectures.

Nearly all models of direction selectivity (DS) in visual cortex are based on feedforward connection schemes, where geniculate input provides all excitatory synaptic input to both pyramidal and inhibitory neurons. Feedforward inhibition then suppresses feedforward excitation for nonoptimal stimuli. Anatomically, however, the majority of asymmetric, excitatory, synaptic contacts onto cortical cells is provided by other cortical neurons, as embodied in the Canonical Microcircuit of Douglas and Martin (1991). In this view, weak geniculate input is strongly amplified in the preferred direction by the action of intracortical excitatory connections, while in the null direction inhibition reduces geniculate-induced excitation. We investigate analytically and through biologically realistic computer simulations the functioning of a cortical network based on massive excitatory, cortico-cortical feedback. The behavior of this network is compared to physiological data as well as to the behavior of a purely feedforward model of DS based on nonlagged input. Our model explains a number of puzzling features of direction selective simple cells, including the small somatic input conductance changes that have been measured experimentally during stimulation in the null direction, and the persistence of DS while fully blocking inhibition in a single cell. Although the operation at the heart of our network is amplification, the network passes the linearity test of (Jagadeesh et al., 1993). We make specific predictions concerning the effect of selective blockade of cortical inhibition on the velocity-response curve.

1. Computer simulations were used to study the effect of voltage-dependent calcium and potassium conductances in the apical dendritic tree of a pyramidal cell on the synaptic efficacy of apical synaptic input. The apical tuft in layers 1 and 2 is the target of feedback projections from other cortical areas. 2. The current, Isoma, flowing into the soma in response to synaptic input was used to assess synaptic efficacy. This measure takes full account of all the relevant nonlinearities in the dendrities and can be used during spiking activity. Isoma emphasizes current flowing in response to synaptic input rather than synaptically induced voltage change. This measure also permits explicit characterization of the input-output relationship of the entire neuron by computing the relationship between presynaptic input and postsynaptic output frequency. 3. Simulations were based on two models. The first was a biophysically detailed 400-compartment model of a morphologically characterized layer 5 pyramidal cell from striate cortex of an adult cat. In this model eight voltage-dependent conductances were incorporated into the somatic membrane to provide the observed firing behavior of a regular spiking cell. The second model was a highly simplified three-compartment equivalent electrical circuit. 4. If the dendritic tree is entirely passive, excitatory synaptic input of the non-N-methyl-D-aspartate (non-NMDA) type to layers 1, 2, and 3 saturate at very moderate input rates, because of the high input impedance of the apical tuft. Layers 1 and 2 together can deliver only 0.25 nA current to the soma. This modest effect is surprising in view of the important afferents that synapse on the apical tuft and is inconsistent with experimental data indicating a more powerful effect. 5. We introduced in a controlled manner a voltage-dependent potassium conductance in the apical tuft, gK, to prevent saturation of the synaptic response. This conductance was designed to linearize the relationship between presynaptic input frequency and the somatic current. We also introduced a voltage-dependent calcium conductance along the apical trunk, gCa, to amplify the apical signal, i.e., the synaptic current reaching the soma. 6. To arrive at a specific relationship between the presynaptic input rate and the somatic current delivered by the synaptic input, we derived the activation curves of gK and gCa either analytically or numerically. The resultant voltage-dependent behavior of both conductances was similar to experimentally measured activation curves.(ABSTRACT TRUNCATED AT 400 WORDS)

It is widely held that visual cortical neurons encode information primarily in their mean firing rates. Some proposals, however, emphasize the information potentially available in the temporal structure of spike trains (Optican and Richmond, 1987; Bialek et al., 1991), in particular with respect to stimulus-related synchronized oscillations in the 30-70 Hz range (Eckhorn et al., 1988; Gray et al., 1989; Kreiter and Singer, 1992) as well as via bursting cells (Cattaneo et al., 1981a; Bonds, 1992). We investigate the temporal fine structure of spike trains recorded in extrastriate area MT of the trained macaque monkey, a region that plays a major role in processing motion information. The data were recorded while the monkey performed a near-threshold direction discrimination task so that both physiological and psychophysical data could be obtained on the same set of trials (Britten et al., 1992). We identify bursting cells and quantify their properties, in particular in relation to the behavior of the animal. We compute the power spectrum and the distribution of interspike intervals (ISIs) associated with individual spike trains from 212 cells, averaging these quantities across similar trials. (1) About 33% of the cells have a relatively flat power spectrum with a dip at low temporal frequencies. We analytically derive the power spectrum of a Poisson process with refractory period and show that it matches the observed spectrum of these cells. (2) About 62% of the cells have a peak in the 20-60 Hz frequency band. In about 10% of all cells, this peak is at least twice the height of its base. The presence of such a peak strongly correlates with a tendency of the cell to respond in bursts, that is, two to four spikes within 2-8 msec. For 93% of cells, the shape of the power spectrum did not change dramatically with stimulus conditions. (3) Both the ISI distribution and the power spectrum of the vast majority of bursting cells are compatible with the notion that these cells fire Poisson-distributed bursts, with a burst-related refractory period. Thus, for our stimulus conditions, no explicitly oscillating neuronal process is required to yield a peak in the power spectrum. (4) We found no statistically significant relationship between the peak in the power spectrum and psychophysical measures of the monkeys' performance on the direction discrimination task.(ABSTRACT TRUNCATED AT 400 WORDS)

How random is the discharge pattern of cortical neurons? We examined recordings from primary visual cortex (V1; Knierim and Van Essen, 1992) and extrastriate cortex (MT; Newsome et al., 1989a) of awake, behaving macaque monkey and compared them to analytical predictions. For nonbursting cells firing at sustained rates up to 300 Hz, we evaluated two indices of firing variability: the ratio of the variance to the mean for the number of action potentials evoked by a constant stimulus, and the rate-normalized coefficient of variation (Cv) of the interspike interval distribution. Firing in virtually all V1 and MT neurons was nearly consistent with a completely random process (e.g., Cv approximately 1). We tried to model this high variability by small, independent, and random EPSPs converging onto a leaky integrate-and-fire neuron (Knight, 1972). Both this and related models predicted very low firing variability (Cv << 1) for realistic EPSP depolarizations and membrane time constants. We also simulated a biophysically very detailed compartmental model of an anatomically reconstructed and physiologically characterized layer V cat pyramidal cell (Douglas et al., 1991) with passive dendrites and active soma. If independent, excitatory synaptic input fired the model cell at the high rates observed in monkey, the Cv and the variability in the number of spikes were both very low, in agreement with the integrate-and-fire models but in strong disagreement with the majority of our monkey data. The simulated cell only produced highly variable firing when Hodgkin-Huxley-like currents (INa and very strong IDR) were placed on distal dendrites. Now the simulated neuron acted more as a millisecond-resolution detector of dendritic spike coincidences than as a temporal integrator. We argue that neurons that act as temporal integrators over many synaptic inputs must fire very regularly. Only in the presence of either fast and strong dendritic nonlinearities or strong synchronization among individual synaptic events will the degree of predicted variability approach that of real cortical neurons.

Dendritic spines are tiny, specialized protoplasmic protuberances that cover the surface of many neurons. First described by Ramon y Cajal(199 1) in light microscopic studies of Golgistained tissue, they are among the most striking subcellular features of many neurons. Spines serve as the major target for excitatory synaptic input onto principal neurons in the hippocampus, the neocortex, and other brain regions. Their intimate association with synaptic traffic suggests some critical role in synaptic transmission and plasticity. We here review experimental data and theoretical models with respect to the putative role of dendritic spines in the induction and retention of synaptic plasticity.

When a typical nerve cell is injected with enough current, it fires a regular stream of action potentials. But cortical cells in vivo usually fire irregularly, reflecting synaptic input from presynaptic cells as well as intrinsic biophysical properties. We have applied the theory of stochastic&#13;\nprocesses to spike trains recorded from cortical neurons (Tuckwell 1989) and find a fundamental contradiction between the large interspike variability observed and the much lower values predicted by well-accepted biophysical models of single cells.

The standard one-dimensional Rall cable model assumes that the electrotonic structure of neurons does not change in response to synaptic input. This model is used in a great number of both theoretical and anatomical-physiological structure-function studies. In particular, the membrane time constant, tau m, the somatic input resistance, Rin, and the electrotonic length are used to characterize single cells. However, these studies do not take into account that neurons are embedded in a network of spontaneously active cells. Synapses from these cells will contribute significantly to the membrane conductance, especially if recent evidence of very high specific membrane resistance, Rm = 100 k omega.cm2, is taken into account. We numerically simulated the electrical behavior of an anatomically reconstructed layer V cortical pyramidal cell receiving input from 4000 excitatory and 1000 inhibitory cells firing spontaneously at 0-7 Hz. We found that, over this range of synaptic background activity, tau m and Rin change by a factor of 10 (80-7 msec, 110-14 M omega) and the electrotonic length of the cell changes by a factor of 3. We show that this significantly changes the response of the cell to temporal desynchronized versus temporal synchronized synaptic input distributed throughout the neuron. Thus, the global activity of the network can control how individual cells perform spatial and temporal integration.

In order to arrive at a quantitative understanding of the dynamics of cortical neuronal networks, we simulated a detailed model of the primary visual pathway of the adult cat. This computer model comprises a 5 degrees x 5 degrees patch of the visual field at a retinal eccentricity of 4.5 degrees and includes 2048 ON- and OFF-center retinal beta-ganglion cells, 8192 geniculate X-cells, and 4096 simple cells in layer IV in area 17. The neurons are implemented as improved integrate-and-fire units. Cortical receptive fields are determined by the pattern of afferent convergence and by inhibitory intracortical connections. Orientation columns are implemented continuously with a realistic receptive field scatter and jitter in the preferred orientations. We first show that realistic ON-OFF-responses, orientation selectivity, velocity low-pass behaviour, null response, and responses to spot stimuli can be obtained with an appropriate alignment of geniculate neurons converging onto the cortical simple cell (Hubel and Wiesel, 1962) and in the absence of intracortical connections. However, the average receptive field elongation (length to width) required to obtain realistic orientation tuning is 4.0, much higher than the average observed elongation. This strongly argues for additional intracortical mechanisms sharpening orientation selectivity. In the second stage, we simulated five different inhibitory intracortical connection patterns (random, local, sparse-local, circular, and cross-orientation) in order to investigate the connection specificity necessary to achieve orientation tuning. Inhibitory connection schemes were superimposed onto Hubel and Wiesel-type receptive fields with an elongation of 1.78. Cross-orientation inhibition gave rise to different horizontal and vertical orientation tuning curves, something not observed experimentally. A combination of two inhibitory schemes, local and circular inhibition (a weak form of cross-orientation inhibition), is in good agreement with observed receptive field properties. The specificity required to establish these connections during development is low. We propose that orientation selectivity is caused by at least three different mechanisms ("eclectic" model): a weak afferent geniculate bias, broadly tuned cross-orientation inhibition, and some iso-orientation inhibition. The most surprising finding is that an isotropic connection scheme, circular inhibition, in which a cell inhibits all of its postsynaptic target cells at a distance of approximately 500 microns, enhances orientation tuning and leads to a significant directional bias. This is caused by the embedding of cortical cells within a columnar structure and does not depend on our specific assumptions.(ABSTRACT TRUNCATED AT 400 WORDS)

We present a biophysical model of electrical and Ca(2+) dynamics following activation of N-methyl-D-aspartate (NMDA) receptors located on a dendritic spine. The model accounts for much of the phenomenology of the induction of long-term potentiation at a Hebbian synapse in hippocampal region CA1. Computer simulations suggested four important functions of spines in this Ca(2+)-dependent synaptic modification: (i) compartmentalizing transient changes in [Ca(2+)] to just those synapses that satisfy the conjunctive requirement for synaptic modification; (ii) isolating the spine head from changes in the [Ca(2+)] at the dendritic shaft; (iii) amplifying the concentration changes at those synapses; and (iv) increasing the voltage dependence of the processes underlying long term potentiation induction. This proposed role of spines in the regulation of Ca(2+) dynamics contrasts with traditional approaches to spine function that have stressed electronic properties. This model can be used to explore the computational implications of Hebbian synapses.

Koch 1990) that outlined a sketch of such a theory. Our aim was not to produce as complete a theory of con-sciousness as possible but to indicate promising lines of experimental work, mainly neurobiological, that might lead eventually toward a solution of the problem. We made the plausible assumption that all forms of con-sciousness (e.g., seeing, thinking, and pain) employ, at bottom, rather similar mechanisms and that if one form were understood, it would be much easier to tackle the others. We then made the personal choice of the mam-malian visual system as the most promising one for an experimental ttack. This choice means that fascinating aspects of the subject, such as volition, intentionality, and self-consciousness, to say nothing of the problem of qualia, have had to be left on one side. We have also not dwelt on unusual psychological states, such as hypnosis, dreaming, lucid dreaming, and sleepwalking, to say nothing of meditative states, since we do not see any special advantage in studying them experimentally at this stage. Our method is to combine what is known experimen-tally from both psychological nd neurobiological ex-periments with plausible theoretical arguments of a general nature. We have not so far attempted etailed computer simulations, ince it is unlikely that we under-stand the details of the system well enough for such modeling to provide decisive answers. In the long run, however, modeling is essential, and even at this stage, well-conceived computer simulations should suggest plausible general principles and thus guide the ex-perimental attack on what is obviously a system of enormous complexity and subtlety. In our previous paper (Crick and Koch 1990), we listed a number of topics and assumptions that we did not intend to discuss. This list will not be repeated here in toto except o say that we have assumed the higher mammals have a form of visual awareness fairly similar to ours, so that we can, with discretion, combine ex-perimental results on humans with those from mon-keys, cats, and so on.

A recent report has provided evidence that there are no significant increases in the neuronal input conductance during the response of cortical cells in cat visual cortex to non-preferred visual stimuli (Douglas et al., 1988). A criticism of experiments of this kind is that changes in the membrane conductance occurring in the dendritic tree may not be visible from electrodes that impale the soma. Our paper describes theoretical and numerical results concerning the visibility of synaptically induced conductance changes from intracellular electrodes, in both ideal and anatomically well-characterized cortical neurons. Based on earlier work by Rall (1967), we here derive theoretical expressions for the change in input conductance at any location in a passive dendritic tree resulting from activation of a single synapse and obtain bounds for the effects of multiple synapses. We find that the conductance change measured at the cell body is always less than the sum of the synaptic conductance changes and that this observed conductance change does not depend on the synaptic reversal potential. For the case of an infinite dendritic cylinder, the change in input resistance due to a single synaptic input decays exponentially with distance of the synapse from the recording site. Numerical simulations of synaptic inputs that change approximately as fast as the membrane time-constant produce an increase in input conductance that is only slightly less visible than that of a constant input. We also compute the changes in somatic input conductance of 2 morphologically identified pyramidal cells from cat visual cortex during activity of a single inhibitory basket cell with known synaptic input locations. We find that the increase in conductance due to the activity of the inhibitory basket cells is clearly visible from the cell body of the pyramidal cells and that a 70% reduction in the amplitude of excitation is associated with at least a 30% increase in somatic input conductance, which would be visible in intracellular recordings. Taken together with the negative experimental evidence of Douglas et al. (1988), our results cast doubt on a large class of models of direction selectivity that rely on synaptically mediated inhibitory conductance increases to veto or block excitatory conductances increases.

Visual awareness is a favorable form of consciousness to study neurobiologically. We propose that it takes two forms: a very fast form, linked to iconic memory, that may be difficult to study; and a somewhat slower one involving visual attention and short-term memory. In the slower form an attentional &#13;\nmechanism transiently binds together all those neurons whose activity relates to the relevant features of a single visual object. We suggest this is done by generating coherent semi-synchronous oscillations, probably in the 40-70 Hz range. These oscillations then activate a transient short-term &#13;\n(working) memory. We outfit several lines of experimental work that might advance the understanding of the neural mechanisms involved. The neural basis of very short-term memory especially needs more experimental study. &#13;\n

Methods in Neuronal is the first technical handbook on computational neuroscience. Written for researchers and theoreticians alike, it outlines methods and techniques used for simulating on digital computers the functional properties of single neurons from synapses, dendrites, single cells; and small invertebrate networks to large scale neural networks in the mammalian nervous system.The use of new experimental tools such as selective staining methods, membrane patch electrodes, voltage and calcium-dependent dyes, and multielectrode recordings, together with the, advent of universally available powerful computing, makes it possible to construct detailed and realistic models of neuronal systems. Methods in Neuronal Modeling addresses such questions as what can and should be simulated and what techniques should be used; what experimental parameters are crucial for such simulations, and whether these models may be verified experimentally.Chapters cover simulation of passive dendritic trees, compartmental models of single cells including neurons with a number of different ionic channels, calcium current dynamics, simulations of small invertebrate networks, simulations of the mammalian cortex, connectionists' models, and the use of parallel computers in modeling neural networks. Although the chapters were written by several authors, they are uniform in structure and notation. Detailed examples are given to clarify the different approaches. Each chapter concludes with a description of the model discussed and the details of its implementation on the computer.Christof Koch is an Assistant Professor of Computation and Neural Systems at the California Institute of Technology. Idan Segev is a Lecturer in Neurobiology at the Institute of Life Science, Hebrew University of Jerusalem. Methods in Neuronal Modeling inaugurates the new series in Computational Neuroscience, edited by Terrence J. Sejnowski and Tomaso Poggio. A Bradford Book.

The authors describe recent developments in the theory of early vision that led from the formulation of the motion problem as an ill-posed one to its solution by minimizing certain 'cost' functions. These cost or energy functions can be mapped onto simple analog and digital resistive networks. The optical flow is computed by injecting currents into resistive networks and recording the resulting stationary voltage distribution at each node. The authors believe that these networks, which they implemented in complementary metal-oxide-semiconductor (CMOS) very-large-scale integrated (VLSI) circuits, represent plausible candidates for biological vision systems.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

The ultimate aim of computational neuroscience is to explain how electrical and chemical signals are used in the brain to represent and process information. This goal is not new, but much has changed in the last decade. More is known now about the brain because of advances in neuroscience, more computing power is available for performing realistic simulations of neural systems, and new insights are available from the study of simplifying models of large networks of neurons. Brain models are being used to connect the microscopic level accessible by molecular and cellular techniques with the systems level accessible by the study of behavior.

Increased levels of intracellular calcium at either pre- or postsynaptic sites are thought to precede changes in synaptic strength. Thus, to induce long-term potentiation in the hippocampus, periods of intense synaptic stimulation would have to transiently raise the levels of cytosolic calcium at postsynaptic sites--dendritic spines in the majority of cases. Since direct experimental verification of this hypothesis is not possible at present, calcium levels have been studied by numerically solving the appropriate electro-diffusion equations for two different postsynaptic structures. Under the assumption that voltage-dependent calcium channels are present on dendritic spines, free intracellular calcium in spines can reach micromolar levels after as few as seven spikes in 20 milliseconds. Moreover, a short, but high-frequency, burst of presynaptic activity is more effective in raising levels of calcium and especially of the calcium-calmodulin complex than sustained low-frequency activity. This behavior is different from that seen at the soma of a typical vertebrate neuron.

The measurement and use of visual motion is one of the most fundamental abilities of biological vision systems, serving many essential functions. For example, a sudden movement in the scene might indicate an approaching predator or a desirable prey. The rapid expansion of features in the visual field can signal an object about to collide with the observer. Discontinuities in motion often occur at the locations of object boundaries and can be used to carve up the scene into distinct objects. Motion signals provide input to centers controlling eye movements, allowing objects of interest to be tracked through the scene. Relative movement can be used to infer the three-dimensional (3-D) structure and motion of object surfaces, and the movement of the observer relative to the scene, allowing biological systems to navigate quickly and efficiently through the environment. More generally, the analysis of visual motion helps us to maintain continuity of our perception of the constantly changing environment around us.

Many problems in early vision can be formulated in terms of minimizing a cost function. Examples are shape from shading, edge detection, motion analysis, structure from motion, and surface interpolation. As shown by Poggio and Koch [Poggio, T. & Koch, C. (1985) Proc. R. Soc. London, Ser. B 226, 303-323], quadratic variational problems, an important subset of early vision tasks, can be "solved" by linear, analog electrical, or chemical networks. However, in the presence of discontinuities, the cost function is nonquadratic, raising the question of designing efficient algorithms for computing the optimal solution. Recently, Hopfield and Tank [Hopfield, J. J. & Tank, D. W. (1985) Biol. Cybern. 52, 141-152] have shown that networks of nonlinear analog "neurons" can be effective in computing the solution of optimization problems. We show how these networks can be generalized to solve the nonconvex energy functionals of early vision. We illustrate this approach by implementing a specific analog network, solving the problem of reconstructing a smooth surface from sparse data while preserving its discontinuities. These results suggest a novel computational strategy for solving early vision problems in both biological and real-time artificial vision systems.

Abstract We outline a theoretical framework that leads from the computational nature of early vision to algorithms for solving them and finally to a specific class of analogue and parallel hardware for the efficient solution of these algorithms. The common computational structure of many early vision problems is that they are mathematically ill-posed in the sense of Hadamard. Regularization analysis can be used to solve them in terms of variational principles of a specific type that enforce constraints derived from a physical analysis of the problem. Studies of human perception may reveal whether principles of a similar type are exploited by biological vision. We also show that the corresponding variational principles can be implemented in a natural way by analogue networks. Specific electrical and chemical networks for localizing edges and computing visual motion are derived. We suggest that local circuits of neurons may exploit this unconventional model of computation.

In a passive dendritic tree, inhibitory synaptic inputs activating ionic conductances with an equilibrium potential near the resting potential can effectively veto excitatory inputs. Analog interactions of this type can be very powerful if the inputs are appropriately timed and occur at certain locations. We examine with computer simulations the precise conditions required for strong and specific interactions in the case of a delta-like ganglion cell of the cat retina. We find some critical conditions to be that (i) the peak inhibitory conductance changes must be sufficiently large (i.e., approximately equal to 50 nS or more), (ii) inhibition must be on the direct path from the location of excitation to the soma, and (iii) the time course of excitation and inhibition must substantially overlap. Analog AND-NOT operations realized by satisfying these conditions may underlie direction selectivity in ganglion cells.

Abstract The electrical properties of a cortical (spiny) pyramidal cell were analysed on the basis of passive cable theory from measurements made on his­tological material (C. Koch, T. Poggio &amp; V. Torre, Phil. Trans. R. Soc. Lond. B 298, 227-264 (1982)). The basis of this analysis is the solution of the cable equation for an arbitrary branched dendritic tree. The con­clusions, however, hold within a wide range of values of electrical par­ameters, provided that the membrane is passive. We determined the potential at the soma as a function of the synaptic input (transient conductance changes) and as a function of the spine neck dimensions, following a suggestion by W. Rall (Brain Inf. Serv. Res. Rep. 3, 13-21 (1974); Studies in neurophysiology (ed. R. Porter), pp. 203–209 (Cambridge University Press, 1978)) that the spine neck might be an important determinant in regulating the efficiency of synapses on spines. From our investigation four major points emerge. (i) Spines may effectively compress the effect of each single excitatory synapse on the soma, mapping a wide range of inputs onto a limited range of outputs (nonlinear saturation). This is also true for very fast transient inputs, in sharp contrast with the case of a synapse on a dendrite. (ii) The somatic depolarization due to an excitatory synapse on a spine is a very sensitive function of the spine neck length and diameter. Thus the spine can effectively control the attenuation of its input via the dimensions of the neck, thereby setting the shape of the resulting saturation curve. There is an optimal neck diameter for which variations of the neck are most effective in controlling the weight of the excitatory spine synapse. For reasonable parameter values this optimal value is consistent with anatomical data. This might be the basic mechanism underlying ultra-short memory, long-term potentiation in the hippo­ campus or learning in the cerebellum. (iii) Spines with shunting inhibitory synapses on them are ineffective in reducing the somatic depolarization due to excitatory inputs on the dendritic shaft or on other spines. Thus isolated inhibitory synapses on a spine are not expected to occur. (iv) The conjunction of an excitatory synapse with a shunting inhibitory synapse on the same spine may result in a time-discrimination circuit with a temporal resolution of around 100 μs.

Abstract The electrical properties of the different anatomical types of retinal ganglion cells in the cat were calculated on the basis of passive cable theory from measurements made on histological material. Standard values for the electrical parameters were assumed (R1 = 70 Ω cm, Cm = 2 μF cm-2, Rm = 2500 Ω cm2). We conclude that these neurons need not be equipotential despite their small dimensions, mainly because of their extensive branching. The interactions between excitation and inhibition when the inhibitory battery is near the resting potential can be strongly nonlinear in these cells. To characterize the different types of ganglion cells in terms of this property we introduce the factor by which the soma depolarization induced by a given excitatory input is reduced by inhibition. In this framework we analyse some of the integrative properties of an arbitrary passive dendritic tree and we then derive the functional properties that are characteristic for the various types of ganglion cells. Our main results are: (i) Nonlinear saturation at the synapses may be made effectively smaller by spreading the same (conductance) input among several subunits on the dendritic field. Subunits are defined as regions of the dendritic field that are somewhat isolated from each other and roughly equipotential within. (ii) Shunting inhibition can specifically veto an excitatory input, if it is located on the direct path to the soma. The F values can then be very high even when the excitatory inputs are much larger than the inhibitory, as long as the absolute value of inhibition is not too small. Inhibition more distal than excitation is much less effective. (iii) Specific branching patterns coupled with suitable distribution of synapses are potentially able to support complex information processing operations on the incoming excitatory and inhibitory signals. The quantitative analysis of the morphology of cat retinal ganglion cells leads to the following specific conclusions: (i) None of the cells examined satisfies Rail’s equivalent cylinder condition. The dendritic tree cannot be satisfactorily approximated by a non-tapering cylinder. (ii) Under the assumption of a passive membrane, the dendritic architecture of the different types of retinal ganglion cells reflects characteristically different electrical properties, which are likely to be relevant for their physiological function and their information processing role: (a) α cells have spatially inhomogeneous electrical properties, with many subunits. Within each subunit nonlinear effects may take place; between subunits good linear summation is expected. F values are relatively low. (b) β cells at small eccentricities have rather homogeneous electrical properties. Even distal inputs are weighted rather uniformly. Electrical inhomogeneities of the a type appear for P cells at larger eccentricities. F values are low. (c) γ-like cells have few subunits, each with high input resistance underlying nonlinear saturation effects possibly related to a sluggish character. F values are high: inhibition of the shunting type can interact in a strongly nonlinear way with excitatory conductance inputs. (d) δ-like cells show many subunits with a high input resistance, covering well the dendritic area. Within each subunit inhibition on the direct path to the soma can specifically veto a more distal excitation. It is conjectured that such a synaptic organization superimposed on the δ cell morphology underlies directional selectivity to motion. (iii) Most of our data refer to steady-state properties. They probably apply, however, to all light evoked signals, since transient inputs with time to peak of 30 ms or more can be treated in terms of steady-state properties of the ganglion cells studied. (iv) All our results are affected only slightly by varying the parameter values within reasonable ranges. If, however, the membrane resistance were very high, all ganglion cells would approach equipotentiality.