Brain Computations and Connectivity

Name: Brain Computations and Connectivity
Brand: OUP eBook
Price: 130.99 EUR
Availability: OnlineOnly

Edmund T. Rolls(Autor*in)

OUP eBook (Verlag)

2. Auflage

Erschienen am 26. Juni 2023

1184 Seiten

E-Book

PDF mit Adobe-DRM

Systemvoraussetzungen

978-0-19-888792-8 (ISBN)

130,99 €inkl. 7% MwSt.

Systemvoraussetzungen

für PDF mit Adobe-DRM

E-Book Einzellizenz

Als Download verfügbar

Beschreibung

Weitere Details

Weitere Ausgaben

Person

Inhalt

Cover
Brain Computations and Connectivity
Copyright
Preface
Contents
1: Introduction
1.1 What and how the brain computes: introduction
1.2 What and how the brain computes: plan of the book
1.3 Neurons in the brain, and their representation in neuronal networks
1.4 A formalism for approaching the operation of single neurons in a network
1.5 Synaptic modification
1.6 Long-term potentiation and long-term depression as models of synaptic modification
1.6.1 Long-Term Potentiation
1.6.2 Long-Term Depression
1.6.3 Spike Timing-Dependent Plasticity
1.7 Information encoding by neurons, and distributed representations
1.7.1 Definitions
1.7.2 Advantages of sparse distributed encoding
1.8 Neuronal network approaches versus connectionism
1.9 Introduction to three neuronal network architectures
1.10 Systems-level analysis of brain function
1.11 Brodmann areas
1.12 Human Connectome Project Multi-Modal Parcellation atlas of the human cortex
1.13 Connectivity of the human brain
1.13.1 Connections analyzed with diffusion tractography
1.13.2 Functional connectivity
1.13.3 Effective connectivity
1.14 Introduction to the fine structure of the cerebral neocortex
1.14.1 The fine structure and connectivity of the neocortex
1.14.2 Excitatory cells and connections
1.14.3 Inhibitory cells and connections
1.14.4 Quantitative aspects of cortical architecture
1.14.5 Functional pathways through the cortical layers
1.14.6 The scale of lateral excitatory and inhibitory effects, and the concept of modules
2: The ventral visual system
2.1 Introduction and overview
2.1.1 Introduction
2.1.2 Overview of what is computed in the ventral visual system
2.1.3 Overview of how computations are performed in the ventral visual system
2.1.4 What is computed in the ventral visual system is unimodal, and is related to other 'what' systems after the inferior temporal visual cortex
2.2 What: V1 - primary visual cortex
2.3 What: V2 and V4 - intermediate processing areas in the ventral visual system
2.4 What: Invariant representations of faces and objects in the inferior temporal visual cortex
2.4.1 Reward value is not represented in the primate ventral visual system
2.4.2 Translation invariant representations
2.4.3 Reduced translation invariance in natural scenes, and the selection of a rewarded object
2.4.4 Size and spatial frequency invariance
2.4.5 Combinations of features in the correct spatial configuration
2.4.6 A view-invariant representation
2.4.7 Learning of new representations in the temporal cortical visual areas
2.4.8 A sparse distributed representation is what is computed in the ventral visual system
2.4.9 Face expression, gesture, and view represented in a population of neurons in the cortex in the superior temporal sulcus
2.4.10 Specialized regions in the temporal cortical visual areas
2.5 The connectivity of the ventral visual pathways in humans
2.5.1 A Ventrolateral Visual Cortical Stream to the inferior temporal visual cortex for object and face representations
2.5.2 A Visual Cortical Stream to the cortex in the inferior bank of the superior temporal sulcus involved in semantic representations
2.5.3 A Visual Cortical Stream to the cortex in the superior bank of the superior temporal sulcus involved in multimodal semantic representations including visual motion, auditory, somatosensory and social information
2.6 How the computations are performed: approaches to invariant object recognition
2.6.1 Feature spaces
2.6.2 Structural descriptions and syntactic pattern recognition
2.6.3 Template matching and the alignment approach
2.6.4 Invertible networks that can reconstruct their inputs
2.6.5 Deep learning
2.6.6 Feature hierarchies and 2D viewbasedobject recognition
2.6.6.1 The feature hierarchy approach to object recognition
2.6.6.2 The Cognitron and Neocognitron
2.7 Hypotheses about how the computations are performed in a feature hierarchy approach to for invariant object recognition
2.8 VisNet: a model of how the computations are performed in the ventral visual system
2.8.1 The architecture of VisNet
2.8.1.1 The memory trace learning rule
2.8.1.2 The network implemented in VisNet
2.8.1.3 Competition and lateral inhibition
2.8.1.4 The input to VisNet
2.8.1.5 Measures for network performance
2.8.2 Initial experiments with VisNet
2.8.2.1 'T','L' and '+' as stimuli: learning translation invariance
2.8.2.2 'T','L', and '+' as stimuli: Optimal network parameters
2.8.2.3 Faces as stimuli: translation invariance
2.8.2.4 Faces as stimuli: view invariance
2.8.3 The optimal parameters for the temporal trace used in the learning rule
2.8.4 Different forms of the trace learning rule, and their relation to error correction and temporal difference learning
2.8.4.1 The modified Hebbian trace rule and its relation to error correction
2.8.4.2 Five forms of error correction learning rule
2.8.4.3 Relationship to temporal difference learning
2.8.4.4 Evaluation of the different training rules
2.8.5 The issue of feature binding, and a solution
2.8.5.1 Syntactic binding of separate neuronal ensembles by synchronization
2.8.5.2 SigmaPineurons
2.8.5.3 Binding of features and their relative spatial position by feature combination neurons
2.8.5.4 Discrimination between stimuli with super- and sub-set feature combinations
2.8.5.5 Feature binding and re-use of feature combinations at different levels of a hierarchical network
2.8.5.6 Feature binding in a hierarchical network with invariant representations of local feature combinations
2.8.5.7 Stimulus generalization to new locations
2.8.5.8 Discussion of feature binding in hierarchical layered networks
2.8.6 Operation in a cluttered environment
2.8.6.1 VisNet simulations with stimuli in cluttered backgrounds
2.8.6.2 Learning invariant representations of an object with multiple objects in the scene and with cluttered backgrounds
2.8.6.3 VisNet simulations with partially occluded stimuli
2.8.7 Learning 3D transforms
2.8.8 Capacity of the architecture, and incorporation of a trace rule into a recurrent architecture with object attractors
2.8.9 Vision in natural scenes - effects of background versus attention
2.8.9.1 Neurophysiology of object selection in the inferior temporal visua lcortex
2.8.9.2 Attention in natural scenes - a computational account
2.8.10 The representation of multiple objects in a scene
2.8.11 Learning invariant representations using spatial continuity: Continuous Spatial Transformation learning
2.8.12 Lighting invariance
2.8.13 Deformation-invariant object recognition
2.8.14 Learning invariant representations of scenes and places
2.8.15 Finding and recognising objects in natural scenes: complementary computations in the dorsal and ventral visual systems
2.8.16 Non-accidental properties, and transform invariant object recognition
2.9 Further approaches to invariant object recognition
2.9.1 Other types of slow learning
2.9.2 HMAX
2.9.3 Minimal recognizable configurations
2.9.4 Hierarchical convolutional deep neural networks
2.9.5 Sigma-Pi synapses
2.9.6 A principal dimensions approach to coding in the inferior temporal visual cortex
2.10 Visuo-spatial scratchpad memory, and change blindness
2.11 Different processes involved in different types of object identification
2.12 Top-down attentional modulation is implemented by biased competition
2.13 Highlights on how the computations are performed in the ventral visual system
3: The dorsal visual system
3.1 Introduction, and overview of the dorsal cortical visual stream
3.2 Global motion in the dorsal visual system
3.3 Invariant object-based motion in the dorsal visual system
3.4 What is computed in the dorsal visual system: visual coordinate transforms
3.4.1 The transform from retinal to head-based coordinates
3.4.2 The transform from head-based to allocentric bearing coordinates
3.4.3 A transform from allocentric bearing coordinates to allocentric spatial view coordinates
3.5 How visual coordinate transforms are computed in the dorsal visual system
3.5.1 Gain modulation
3.5.2 Mechanisms of gain modulation using a trace learning rule
3.5.3 Gain modulation by eye position to produce a head-centered representation in Layer 1 of VisNetCT
3.5.4 Gain modulation by head direction to produce an allocentric bearing to a landmark in Layer 2 of VisNetCT
3.5.5 Gain modulation by place to produce an allocentric spatial view representation in Layer 3 of VisNetCT
3.5.6 The utility of the coordinate transforms in the dorsal visual system
3.6 The human Dorsal Visual Cortical Stream for visual motion leading to the intraparietal visual areas, and then to parietal area 7 regions for actions in space
3.6.1 Dorsal stream visual division regions
3.6.2 MT+ complex regions (FST, LO1, LO2, LO3, MST, MT, PH, V3CD and V4t)
3.6.3 Intraparietal sulcus posterior parietal cortex, regions (AIP, LIPd, LIPv, MIP, VIP
with IP0, IP1 and IP2)
3.6.4 Area 7 regions
4: The taste and flavor system
4.1 Introduction and overview
4.1.1 Introduction
4.1.2 Overview of what is computed in the taste and flavor system
4.1.3 Overview of how computations are performed in the taste and flavor system
4.2 Taste and related pathways: what is computed
4.2.1 Hierarchically organised anatomical pathways
4.2.2 Taste neuronal tuning become more selective through the taste hierarchy
4.2.3 The primary, insular, taste cortex represents what taste is present and its intensity
4.2.4 The secondary, orbitofrontal, taste cortex, and its representation of the reward value and pleasantness of taste
4.2.5 Sensory-specific satiety is computed in the orbitofrontal cortex
4.2.6 Oral texture is represented in the primary and secondary taste cortex: viscosity and fat texture
4.2.7 Vision and olfaction converge using associative learning with taste to represent flavor in the secondary but not primary taste cortex
4.2.8 Top-down attention and cognition can modulate taste and flavor re presentations in the taste cortical areas
4.2.9 The tertiary taste cortex in the anterior cingulate cortex provides the rewards for action-reward outcome learning
4.2.10 Taste, oral texture and flavor provide the rewards for eating, and the gut provides satiety signals
4.3 Taste and related pathways: how the computations are performed
4.3.1 Increased selectivity of taste and flavor neurons through the hierarchy by competitive learning and convergence
4.3.2 Pattern association learning of associations of visual and olfactory stimuli with taste
4.3.3 Rule-based reversal of visual to taste associations in the orbitofrontal cortex
4.3.4 Sensory-specific satiety is implemented by adaptation of synapses onto orbitofrontal cortex neurons
4.3.5 Top-down cognitive and attentional modulation is implemented by biased activation
5: The olfactory system
5.1 Introduction
5.1.1 Overview of what is computed in the olfactory system
5.1.2 Overview of how the computations are performed in the olfactory system
5.2 What is computed in the olfactory system
5.2.1 1000 gene-encodedol factory receptor types, and 1000 corresponding glomerulus types in the olfactory bulb
5.2.2 The primary olfactory, pyriform, cortex: olfactory feature combinations are what is represented
5.2.3 Orbitofrontal cortex: olfactory neuronal response selectivity
5.2.4 Orbitofrontal cortex: olfactory to taste convergence
5.2.5 Orbitofrontal cortex: olfactory to taste association learning and reversal
5.2.6 Orbitofrontal cortex: olfactory reward value is represented
5.2.7 Cognitive influences on olfactory representations in the orbitofrontal cortex
5.3 How computations are performed in the olfactory system
5.3.1 Olfactory receptors, and the olfactory bulb
5.3.2 Olfactory (pyriform) cortex
5.3.3 Orbitofrontal cortex
6: The somatosensory system
6.1 What is computed in the somatosensory system
6.1.1 The receptors and periphery
6.1.2 The anterior somatosensory cortex, areas 1, 2, 3a, and 3b, in the anterior parietal cortex
6.1.3 The ventral somatosensory stream: areas S2 and PV, in the lateral parietal cortex
6.1.4 The dorsal somatosensory stream to area 5 and then 7b, in the posterior parietal cortex
6.1.5 Somatosensory representations in the insula
6.1.6 Somatosensory and temperature inputs to the orbitofrontal cortex, affective value, pleasant touch, and pain
6.1.7 Decision-making in the somatosensory system
6.1.8 Somatosensory cortical regions and connectivity in humans
6.1.8.1 A ventral somatosensory stream via the insula to the inferior parietal cortex
6.1.8.2 A dorsal somatosensory stream
6.1.8.3 Taste cortical regions
6.1.8.4 Connectivity of somatosensory to motor and premotor frontal cortical regions
6.2 How computations are performed in the somatosensory system
6.2.1 Hierarchical computation in the somatosensory system
6.2.2 Computations for pleasant touch and pain
6.2.3 The mechanisms for somatosensory decisionmaking
7: The auditory system
7.1 Introduction, and overview of computations in the auditory system
7.2 Auditory Localization
7.3 Ventral and dorsal cortical auditory pathways
7.4 The ventral cortical auditory stream
7.5 The dorsal cortical auditory stream
7.6 Auditory cortical regions and connectivity in humans
7.6.1 Early Auditory cortical regions
7.6.2 Ventral auditory cortical streams
7.6.3 Dorsal auditory cortical streams
7.6.4 Other auditory system cortical connectivities
7.7 How the computations are performed in the auditory system
8: The temporal cortex
8.1 Introduction and overview
8.2 Middle temporal gyrus and face expression and gesture
8.3 Semantic representations in the temporal lobe neocortex
8.3.1 Neurophysiology of the medial temporal lobe, including concept cells
8.3.2 Neuropsychology
8.3.3 Functional neuroimaging
8.3.4 Brain stimulation
8.4 Connectivity and functions of the human temporal lobe regions related to semantics
8.4.1 Group 1 semantic regions that include regions in the ventral bank of the Superior Temporal Sulcus
8.4.2 Group 3 semantic regions that include regions in the dorsal bank of the Superior Temporal Sulcus
8.5 The mechanisms for semantic learning in the human anterior temporal lobe
9: The hippocampus, memory, and spatial function
9.1 Introduction and overview
9.1.1 Overview of what is computed by the hippocampal system
9.1.2 Overview of how the computations are performed by the hippocampal system
9.2 What is computed in the hippocampus
9.2.1 Systems-level anatomy
9.2.2 Evidence from the effects of damage to the hippocampus
9.2.3 Episodic memories need to be recalled from the hippocampus, and can be used to help build neocortical semantic memories
9.2.4 Systems-level neurophysiology of the primate including human hippocampus
9.2.4.1 Spatial view neurons
Spatial view cells are allocentric
Spatial view cells and place cells in the primate hippocampus
Spatial view cells during active locomotion in the primate hippocampus and parahippocampal gyrus
Spatial view cells are selective for spatial view much more than for place or head direction or eye position
Spatial visual responses during navigation in a virtual environment
Idiothetic (self-motion) update of spatial view cells
9.2.4.2 Whole body motion or speed neurons
9.2.4.3 Spatial view grid cells in the primate entorhinal cortex
9.2.4.4 Spatial view associated with object neurons in association memory
9.2.4.5 Spatial view associated with reward neurons
9.2.5 Spatial view cells in primates including humans, and foveal vision
9.2.6 Head direction cells in the presubiculum
9.2.7 Perirhinal cortex, recognition memory, and long-term familiarity memory
9.2.8 Connectivity of the human hippocampal system
9.2.8.1 A ventromedial visual cortical stream to the hippocampus for visual scene representations
9.2.8.2 The roles of the parietal cortex in the idiothetic update of hippocampal and para hippocampal spatial view cells and scenere presentations
9.2.8.3 Ventrolateral visual stream 'What' inputs to the human hippocampal system
9.2.8.4 Reward value / emotionrelatedinputs to the human hippocampus
9.2.8.5 Human hippocampal system and orbitofrontal / ventromedial prefrontal cortex connectivity involved in memory consolidation
9.3 How computations are performed in the hippocampal system
9.3.1 Historical development of the theory of the hippocampus
9.3.2 Hippocampal circuitry
9.3.3 Medial entorhinal cortex, spatial processing streams, and grid cells
9.3.4 Lateral entorhinal cortex, object processing streams, and the generation of time cells in the hippocampus
9.3.4.1 The theory of the generation of time in the brain, and of hippocampal time cells
9.3.4.2 The model of lateral entorhinal cortex temporal cells and hippocampal time cells
9.3.4.3 Operation of the entorhinalhippocampaltime generation network
9.3.4.4 Forward and reverse replay of temporal sequences by the time generation system
9.3.5 CA3 as an autoassociation memory
9.3.5.1 Arbitrary associations, and pattern completion in recall
CA3 recurrent collaterals form a single attractor network
Storage capacity
Recall and Completion
CA3 as a short term attractor memory
9.3.5.2 Continuous, spatial, patterns and CA3 representations: hippocampal charts
9.3.5.3 The dynamics of the recurrent network
9.3.5.4 Memory for sequences
9.3.5.5 The dilution of the CA3 recurrent collateral connectivity enhances memory storage capacity and pattern completion
9.3.5.6 Noise and stability produced by the diluted connectivity and the graded firing rates in the CA3CA3attractor network
9.3.5.7 Pattern separation of CA3 cell populations encoding different memories
9.3.5.8 Mossy fibre inputs to the CA3 cells
9.3.5.9 Perforant path inputs to CA3 cells initiate recall in CA3 and contribute to generalization
9.3.6 Dentate granule cells
9.3.6.1 The dentate cells as a competitive network
9.3.6.2 Separation of overlapping inputs
9.3.6.3 Low contact probability of dentate with CA3 neurons
9.3.6.4 Dentate mossy fibre synapses to CA3 cells are strong
9.3.6.5 Expansion recoding
9.3.6.6 Neurogenesis of dentate granule cells
9.3.7 CA1 cells
9.3.7.1 Associative retrieval at the CA3 to CA1 (Schaffer collateral) synapses
9.3.7.2 Recoding in CA1 to facilitate retrieval to the neocortex
9.3.7.3 CA1 inputs from CA3 vs direct entorhinal inputs
9.3.7.4 CA1 time cells and sequence memory for objects and odors
9.3.7.5 Subcortical, vs neocortical backprojection, outputs from CA1
9.3.8 Backprojections to the neocortex, episodic memory recall, and consolidation
9.3.9 Backprojections to the neocortex - quantitative aspects
9.3.10 Simulations of hippocampal operation
9.3.11 The learning of spatial view and place cell representations from visual inputs
9.3.12 Linking the inferior temporal visual cortex to spatial view and place cell representations
9.3.13 A scientific theory of the art of memory: scientia artis memoriae
9.3.14 How navigation is performed
9.3.14.1 Navigation using a hippocampal allocentric Euclidean cognitive map
9.3.14.2 Navigation using an entorhinal cortex goal vector system
9.3.14.3 Transforms between allocentric and egocentric representations
9.3.15 Navigational computations using neuron types found in primates including humans
9.3.15.1 Navigation using spatial view cells
9.3.15.2 Navigation using allocentric bearing to a landmark cells
9.3.15.3 Navigation using combinations of allocentric bearing to a landmark cells, and triangulation
9.3.15.4 Idiothetic navigation in primates
9.3.15.5 Conclusions on navigation in primates including humans
9.3.15.6 Synthesis: different functions of the parahippocampal scene area and hippocampus in navigation in primates including humans
9.4 Tests of the theory of hippocampal cortex operation
9.4.1 Dentate gyrus (DG) subregion of the hippocampus
9.4.1.1 Spatial pattern separation
9.4.1.2 Pattern separation and neurogenesis
9.4.1.3 Encoding versus retrieval
9.4.2 CA3 subregion of the hippocampus
9.4.2.1 Rapid encoding of new information
9.4.2.2 The types of information associated in CA3
9.4.2.3 Acquisition/encoding associated with multiple trials
9.4.2.4 Pattern completion in CA3
9.4.2.5 Recall and arbitrary associations in CA3
9.4.2.6 Functional analysis of mossy fibre vs direct perforant path input into CA3
9.4.2.7 Orthogonal representations in CA3
9.4.2.8 Shorttermmemory, path integration, and CA3
9.4.2.9 Cholinergic modulation of CA3 towards storage vs recall including pattern completion
9.4.3 CA1 subregion of the hippocampus
9.4.3.1 Sequence memory and CA1
9.4.3.2 Associations across time, and CA1
9.4.3.3 Temporal order memory, and CA1
9.5 Comparison of spatial processing and computations in primates including humans vs rodents
9.5.1 Similarities and differences between the spatial representations in primates and rodents
9.5.2 Hippocampal computational similarities and differences between primates and rodents
9.6 Synthesis: the hippocampus: memory, navigation, or both?
9.7 Comparison with other theories of hippocampal function
10: The parietal cortex, spatial functions, and navigation
10.1 Introduction and overview
10.1.1 Overview of what is computed in the parietal cortex
10.1.2 Overview of how the computations are performed in the parietal cortex
10.2 Inferior parietal cortex somatosensory stream, PF regions
10.3 Intraparietal sulcus posterior parietal cortex, regions AIP, LIPd, LIPv, MIP, VIP, IP0, IP1, and IP2
10.4 Posterior superior parietal cortex, regions 7AL, 7Am, 7PC, 7PL, and 7Pm
10.5 Inferior parietal cortex, visual regions PFm, PGi, PGs and PGp
10.5.1 Region PGi
10.5.2 Region PGs
10.5.3 Region PFm
10.5.4 Region PGp
10.6 Navigation: What computations are performed in the parietal and related cortex
10.7 How the computations are performed in the parietal cortex
11: The orbitofrontal cortex, amygdala, reward value, emotion, and decision-making
11.1 Introduction and overview
11.1.1 Introduction
11.1.2 Overview of what is computed in the orbitofrontal cortex
11.1.3 Overview of how the computations are performed by the orbitofrontal cortex
11.2 The topology and connections of the orbitofrontal cortex
11.2.1 Inputs to the orbitofrontal cortex
11.2.2 Outputs of the orbitofrontal cortex
11.3 What is computed in the orbitofrontal cortex
11.3.1 The orbitofrontal cortex represents reward value
11.3.1.1 Orbitofrontal cortex representations of the reward value for taste, olfaction, oral texture, and touch
11.3.1.2 The expected value of visual stimuli is represented in the orbitofrontal cortex, and can be reversed in one trial
Visual reversal to compute expected value can be rapid and rule-based in primates including humans
Visual stimulus-selective expected value neurons
Devaluation shows that orbitofrontal cortex visual neurons represent expected value
11.3.2 Neuroeconomic value is represented in the orbitofrontal cortex
11.3.3 A representation of face and voice expression and other socially relevant stimuli in the orbitofrontal cortex
11.3.4 Negative reward prediction error neurons in the orbitofrontal cortex, and visual stimulus-reinforcer association learning and reversal during decisionmaking
11.3.5 The human medial orbitofrontal cortex represents rewards, and the lateral orbitofrontal cortex nonrewardand punishers
11.3.6 Decision-making in the orbitofrontal / ventromedial prefrontal cortex
11.3.7 The ventromedial prefrontal cortex and memory
11.3.8 The orbitofrontal cortex and emotion
11.3.9 Emotional orbitofrontal vs rational routes to action
11.3.9.1 Some of the different routes to action produced by emotion-related stimuli
11.3.9.2 Examples of some complex behaviors that may be performed implicitly
11.3.9.3 A reasoning, rational, route to action
11.3.9.4 The Selfish Gene vs The Selfish Phenotype
11.3.9.5 Decisionmakingbetween the implicit and explicit systems
11.3.9.6 The emotional vs rational brain systems, and the strength of emotions in humans
11.3.9.7 The emotional systems in animals, and welfare
11.3.9.8 The orbitofrontal cortex and addiction
11.3.10 The connectivity of the human orbitofrontal cortex, and its relation to function
11.3.10.1 Human medial orbitofrontal cortex connectivity
11.3.10.2 Human lateral orbitofrontal cortex connectivity
11.3.10.3 Connectivity of the orbitofrontal cortex to the hippocampal memory system, and memory consolidation
11.3.10.4 Connectivity of the orbitofrontal cortex to the cholinergic basal forebrain nucleus of Meynert, and memory consolidation
11.3.10.5 Connectivity of the orbitofrontal cortex to the supracallosal anterior cingulate cortex for actionoutcomelearning
11.3.10.6 Connectivity of the orbitofrontal cortex to the basal ganglia and dopaminergic system
11.3.11 Mental problems associated with the orbitofrontal cortex
11.4 What is computed in the amygdala for emotion
11.4.1 Overview of the functions of the amygdala in emotion
11.4.2 The amygdala and the associative processes involved in emotion-related learning
11.4.3 Connections of the amygdala
11.4.4 Effects of amygdala lesions
11.4.5 Amygdala lesions in primates
11.4.6 Amygdala lesions in rats
11.4.7 Neuronal activity in the primate amygdala to reinforcing stimuli
11.4.8 Neuronal responses in the amygdala to faces
11.4.9 Evidence from humans
11.4.10 Connectivity of the human amygdala
11.5 How the computations are performed in the orbitofrontal cortex
11.5.1 Decision-making in attractor networks in the brain
11.5.1.1 An attractor decision-making network
11.5.1.2 An integrate-and-fire implementation of the attractor network for probabilistic decision-making
11.5.2 Analyses of reward-related decision-making mechanisms in the orbitofrontal cortex
11.5.2.1 Neuronal responses on difficult vs easy trials, and decision confidence
11.5.2.2 Decision times of the neuronal responses
11.5.2.3 Percentage correct
11.5.2.4 Prediction of the BOLD signals on difficult vs easy decisionmaking trials
11.5.3 A neurophysiological and computational basis for stimulus-reinforcer association learning and reversal in the orbitofrontal cortex
11.5.4 A theory and model of nonrewardneural mechanisms in the orbitofrontal cortex
11.6 Highlights: what are the special computational roles of the orbitofrontal cortex in reward, emotion, and decision-making?
12: The cingulate cortex
12.1 Introduction to and overview of the cingulate cortex
12.1.1 Introduction
12.1.2 Overview of what is computed in the cingulate cortex
12.1.3 Overview of how the computations are performed by the cingulate cortex
12.2 Anterior cingulate cortex
12.2.1 Anterior cingulate cortex anatomy and connections in primates
12.2.2 Anterior cingulate cortex: A framework
12.2.3 Anterior cingulate cortex and action-outcome representations
12.2.4 Anterior cingulate cortex lesion effects
12.2.5 Anterior cingulate cortex and ventromedial prefrontal cortex connectivity and functions in humans
12.2.5.1 Ventromedial prefrontal cortex connectivity
12.2.5.2 Pregenual anterior cingulate cortex connectivity
12.2.5.3 Subgenual anterior cingulate cortex area 25
12.2.5.4 Supracallosal anterior cingulate cortex
12.2.6 Pregenual anterior cingulate representations of reward value, and supracallosal anterior cingulate representations of punishers and non-reward
12.2.7 The human supracallosal anterior cingulate cortex, dACC, and action-outcome learning
12.2.8 Reward value outputs from the orbitofrontal and pregenual anterior cortex, and vmPFC, to the hippocampal memory system
12.2.9 The pregenual anterior cingulate cortex has connectivity with the septal cholinergic system that is involved in memory consolidation
12.2.10 Reward value outputs from the orbitofrontal and pregenual anterior cortex, and vmPFC, to the hippocampal system to provide the goals for navigation
12.2.11 Subgenual cingulate cortex
12.3 Posterior cingulate cortex
12.3.1 Introduction and overview
12.3.2 Postero-ventral posterior cingulate and medial parietal regions 31pd, 31pv, d23ab, v23ab and 7m, and their relation to episodic memory
12.3.3 Antero-dorsal Posterior Cingulate Division regions 23d, 31a, PCV
and RSC, POS2, and POS1
and their relation to navigation and executive function
12.3.4 Dorsal Visual Transitional area and ProStriate region: the retrosplenial scene area
12.4 Mid-cingulate cortex, the cingulate motor area, and action-outcome learning
12.5 How the computations are performed by the cingulate cortex
12.5.1 The anterior cingulate cortex and emotion
12.5.2 Action-outcome learning in the supracallosal anterior cingulate cortex (dACC)
12.5.3 Connectivity of the posterior cingulate cortex with the hippocampal memory system
12.6 Synthesis and conclusions
13: The prefrontal cortex
13.1 Introduction and overview
13.2 Divisions of the lateral prefrontal cortex
13.2.1 The dorsolateral prefrontal cortex
13.2.2 The caudal prefrontal cortex
13.2.3 The ventrolateral prefrontal cortex
13.3 The connectivity and computational organisation of the human prefrontal cortex
13.3.1 Inferior frontal gyrus
13.3.2 Dorsolateral prefrontal cortex division
13.4 The lateral prefrontal cortex and top-down attention
13.5 The frontal pole cortex
13.6 How the computations are performed in the prefrontal cortex
13.6.1 Cortical short-term memory systems and attractor networks
13.6.2 Prefrontal cortex short-term memory networks, and their relation to temporal and parietal perceptual networks
13.6.2.1 A prefrontal shorttermmemory network supporting a temporal lobe perceptual system
13.6.2.2 Computational details of the prefrontal-inferior temporal model of short-term memory
13.6.3 Mapping from one representation to another in short-term memory
13.6.4 The mechanisms of top-down attention
13.6.5 Computational necessity for a separate, prefrontal cortex, short-term memory system
13.6.6 Synaptic modification is needed to set up but not to reuse short-term memory systems
13.6.7 Sequence memory
13.6.8 Working memory, and planning
14: Language and syntax in the brain
14.1 Introduction and overview
14.1.1 Introduction
14.1.2 Overview
14.2 What is computed in different brain systems to implement language
14.2.1 The Wernicke-Lichtheim-Geschwind hypothesis
14.2.2 The dual-stream hypothesis of speech comprehension
14.2.3 Reading requires different brain systems to hearing speech
14.2.4 Semantic representations
14.2.5 Syntactic processing
14.2.6 The parietal cortex: supramarginal and angular gyri
14.3 Cortical regions for language and their connectivity in humans
14.3.1 A semantic system that includes the inferior bank of the superior temporal sulcus including object representations
14.3.2 A semantic system that includes the superior bank of the superior temporal sulcus including visual motion, auditory, somatosensory and social information
14.3.3 Multimodal semantic representations
14.3.4 Broca's area and related regions (TGv 44 45 47l SFL 55b)
14.4 Hypotheses about how semantic representations are computed
14.5 A neurodynamical hypothesis about how syntax is computed
14.5.1 Binding by synchrony?
14.5.2 Syntax using a place code
14.5.3 Temporal trajectories through a state space of attractors
14.5.4 Hypotheses about the implementation of language in the cerebral cortex
14.5.5 Tests of the hypotheses - a model
14.5.5.1 An integrate-and-fire network with three attractor network modules connected by stronger forward than backward connections
14.5.5.2 The operation of a single attractor network module
14.5.5.3 Spike frequency adaptation mechanism
14.5.6 Tests of the hypotheses - findings with the model
14.5.6.1 A production system
14.5.6.2 A decoding system
14.5.7 Evaluation of the hypotheses
14.5.8 Further approaches
15: The motor cortical areas
15.1 Introduction and overview
15.2 What is computed in different cortical motor-related areas
15.2.1 Ventral parietal and ventral premotor cortex F4
15.2.2 Superior parietal areas with activity related to reaching
15.2.3 Inferior parietal areas with activity related to grasping, and ventral premotor cortex F5
15.3 The mirror neuron system
15.4 How the computations are performed in motor cortical and related areas
16: The basal ganglia
16.1 Introduction and overview
16.2 Systems-level architecture of the basal ganglia
16.3 What computations are performed by the basal ganglia?
16.3.1 Effects of striatal lesions
16.3.1.1 Ventral striatum including nucleus accumbens
16.3.1.2 Dorsal striatum
16.3.2 Neuronal activity in different parts of the striatum
16.3.2.1 Ventral striatum
16.3.2.2 Tail of the caudate nucleus, and posteroventral putamen
16.3.2.3 Postero-ventral putamen
16.3.2.4 Head of the caudate nucleus
16.3.2.5 Anterior putamen
16.4 How do the basal ganglia perform their computations?
16.4.1 Interaction between neurons and selection of output
16.4.2 Convergence within the basal ganglia, useful for stimulus-response habit learning
16.4.3 Dopamine as a reward prediction error signal for reinforcement learning in the striatum
16.5 Comparison of computations for selection in the basal ganglia and cerebral cortex
17: Cerebellar cortex
17.1 Introduction
17.2 Architecture of the cerebellum
17.2.1 The connections of the parallel fibres onto the Purkinje cells
17.2.2 The climbing fibre input to the Purkinje cell
17.2.3 The mossy fibre to granule cell connectivity
17.3 Modifiable synapses of parallel fibres onto Purkinje cell dendrites
17.4 The cerebellar cortex as a perceptron
17.5 Cognitive functions of the cerebellum
17.5.1 Anatomical connections from most neocortical regions
17.5.2 Functional connectivity of different cortical systems with different parts of the cerebellum
17.5.3 Activation of different cerebellar cortical regions in different tasks
17.5.4 Damage to different parts of the cerebellum can produce different cognitive, emotional, and motor impairments
17.5.5 Neocortical-cerebellar cortical computations for cognition
17.6 Highlights: differences between cerebral and cerebellar cortex microcircuitry
18: Cortical attractor dynamics and connectivity, stochasticity, psychiatric disorders, and aging
18.1 Introduction and overview
18.1.1 Introduction
18.1.2 Overview
18.2 The noisy cortex: stochastic dynamics, decisions, and memory
18.2.1 Reasons why the brain is inherently noisy and stochastic
18.2.2 Attractor networks, energy landscapes, and stochastic neurodynamics
18.2.3 A multistable system with noise
18.2.4 Stochastic dynamics and the stability of short-term memory
18.2.4.1 Analysis of the stability of shorttermmemory
18.2.4.2 Stability and noise in a model of shorttermmemory
18.2.5 Stochastic dynamics in decision-making,and the evolutionary utility of probabilistic choice
18.2.6 Selection between conscious vs unconscious decision-making, and free will
18.2.7 Stochastic dynamics and creative thought
18.2.8 Stochastic dynamics and unpredictable behavior
18.3 Attractor dynamics and schizophrenia
18.3.1 Introduction
18.3.2 A dynamical systems hypothesis of the symptoms of schizophrenia
18.3.3 Reduced functional connectivity of some brain regions in schizophrenia
18.3.4 Beyond the disconnectivity hypothesis of schizophrenia: reduced forward but not backward connectivity
18.4 Attractor dynamics and obsessive-compulsive disorder
18.4.1 Introduction
18.4.2 A hypothesis about the increased stability of attractor networks and the symptoms of obsessive-compulsive disorder
18.4.3 Glutamate and increased depth of the basins of attraction of attractor networks
18.5 Depression and attractor dynamics
18.5.1 Introduction
18.5.2 A non-reward attractor theory of depression
18.5.3 The orbitofrontal cortex, and the theory of depression
18.5.4 Altered connectivity of the orbitofrontal cortex in depression
18.5.4.1 Precuneus: higher connectivity with the lateral orbitofrontal cortex
18.5.4.2 Parahippocampal gyrus / medial temporal lobe memory system, and temporal lobe visual cortex: lower connectivity with the medial orbitofrontal cortex
18.5.4.3 Posterior cingulate cortex: higher functional connectivity with the lateral orbitofrontal cortex in depression
18.5.4.4 Anterior cingulate cortex: higher connectivity with the orbitofrontal cortex in depression
18.5.4.5 Inferior frontal gyrus: increased connectivity with the lateral orbitofrontal cortex in depression
18.5.4.6 Amygdala: reduced connectivity with the orbitofrontal cortex in depression
18.5.4.7 Sleep, depression, and increased lateral orbitofrontal cortex connectivity
18.5.4.8 Effective connectivity in depression
18.5.5 Increased activations to non-reward of the lateral orbitofrontal cortex, and decreased sensitivity to reward of the medial orbitofrontal cortex, are related to depression scores
18.5.6 Implications, and possible treatments, and subtypes of depression
18.5.7 Mania and bipolar disorder
18.6 Attractor stochastic dynamics, aging, and memory
18.6.1 NMDA receptor hypofunction
18.6.2 Dopamine and norepinephrine
18.6.3 Impaired synaptic modification
18.6.4 Cholinergic function and memory
18.6.4.1 Cholinergic function, memory, and aging
18.6.4.2 The responses of basal magnocellular forebrain nuclei cholinergic neurons
18.6.4.3 Basal magnocellular forebrain nuclei cholinergic neurons as a cortical strobe to facilitate memory
18.6.4.4 Basal magnocellular forebrain nuclei cholinergic neurons reduce cortical adaptation
18.7 High blood pressure, reduced hippocampal functional connectivity, and impaired memory
18.8 Brain development, and structural differences in the brain
19: Computations by different types of brain, and by artificial neural systems
19.1 Introduction and overview
19.2 Computations that combine different computational systems in the brain to produce behavior
19.3 Brain computation compared to computation on a digital computer
19.4 A comparison of brain computation with learning in artificial deep learning networks using error backpropagation
19.5 Reinforcement Learning
19.6 Levels of explanation, and the mindbrainproblem
19.6.1 A levels of explanation theory of causality, and the relation between the mind and the brain
19.6.2 Downward or Upward Causality?
19.6.3 Consciousness - a Higher Order Syntactic Thought theory
19.6.4 Levels of explanation, and levels of investigation
19.7 Biologically plausible computation in the brain: a grand unifying theory?
19.8 Brain-Inspired Intelligence
19.9 Brain-Inspired Medicine
19.9.1 Computational psychiatry and neurology
19.9.2 Reward systems in the brain, and their application to understanding food intake control and obesity
19.9.3 Multiple Routes to Action
19.10 Primates including humans have a different systems-level organisation of many brain systems compared to rodents
19.10.1 The visual system
19.10.2 The taste system
19.10.3 The olfactory system
19.10.4 The somatosensory system
19.10.5 The auditory system
19.10.6 The hippocampal system, memory, and navigation
19.10.7 The orbitofrontal cortex and amygdala
19.10.8 The cingulate cortex
19.10.9 The motor system
19.10.10 Language
Appendix 1: Introduction to linear algebra for neural networks
A.1 Vectors
A.1.1 The inner or dot product of two vectors
A.1.2 The length of a vector
A.1.3 Normalizing the length of a vector
A.1.4 The angle between two vectors: the normalized dot product
A.1.5 The outer product of two vectors
A.1.6 Linear and non-linear systems
A.1.7 Linear combinations of vectors, linear independence, and linear separability
A.2 Application to understanding simple neural networks
A.2.1 Capability and limitations of singlelayernetworks: linear separability and capacity
A.2.2 Non-linear networks: neurons with non-linear activation functions
A.2.3 Non-linear networks: neurons with non-linear activations
Appendix 2: Neuronal network models
B.1 Introduction
B.2 Pattern association memory
B.2.1 Architecture and operation
B.2.1.1 Learning
B.2.1.2 Recall
B.2.2 A simple model
B.2.3 The vector interpretation
B.2.4 Properties
B.2.4.1 Generalization
B.2.4.2 Graceful degradation or fault tolerance
B.2.4.3 The importance of distributed representations for pattern associators
B.2.5 Prototype extraction, extraction of central tendency, and noise reduction
B.2.6 Speed
B.2.7 Local learning rule
B.2.7.1 Capacity
B.2.7.2 Interference
B.2.7.3 Expansion recoding
B.2.8 Implications of different types of coding for storage in pattern associators
B.3 Autoassociation or attractor memory
B.3.1 Architecture and operation
B.3.1.1 Learning
B.3.1.2 Recall
B.3.2 Introduction to the analysis of the operation of autoassociation networks
B.3.3 Properties
B.3.3.1 Completion
B.3.3.2 Generalization
B.3.3.3 Graceful degradation or fault tolerance
B.3.3.4 Prototype extraction, extraction of central tendency, and noise reduction
B.3.3.5 Speed
B.3.3.6 Local learning rule
B.3.3.7 Capacity
B.3.3.8 Context
B.3.3.9 Mixture states
B.3.3.10 Memory for sequences
B.3.4 Diluted connectivity and the storage capacity of attractor networks
B.3.4.1 The autoassociative or attractor network architecture being studied
B.3.4.2 The storage capacity of attractor networks with diluted connectivity
B.3.4.3 The network simulated
B.3.4.4 The effects of diluted connectivity on the storage capacity of attractor networks
B.3.4.5 Synthesis of the effects of diluted connectivity in attractor networks
B.3.5 Use of autoassociation networks in the brain
B.4 Competitive networks, including self-organizing maps
B.4.1 Function
B.4.2 Architecture and algorithm
B.4.2.1 Architecture
B.4.2.2 Algorithm
B.4.3 Properties
B.4.3.1 Feature discovery by self-organization
B.4.3.2 Removal of redundancy
B.4.3.3 Orthogonalization and categorization
B.4.3.4 Sparsification
B.4.3.5 Capacity
B.4.3.6 Separation of non-linearly separable patterns
B.4.3.7 Stability
B.4.3.8 Frequency of presentation
B.4.3.9 Comparison to principal component analysis (PCA) and cluster analysis
B.4.4 Utility of competitive networks in information processing by the brain
B.4.4.1 Feature analysis and preprocessing
B.4.4.2 Removal of redundancy
B.4.4.3 Orthogonalization
B.4.4.4 Sparsification
B.4.4.5 Brain systems in which competitive networks may be used for orthogonalization and sparsification
B.4.5 Guidance of competitive learning
B.4.6 Topographic map formation
B.4.7 Invariance learning by competitive networks
B.4.8 Radial Basis Function networks
B.4.9 Further details of the algorithms used in competitive networks
B.4.9.1 Normalization of the inputs
B.4.9.2 Normalization of the length of the synaptic weight vector on each dendrite
B.4.9.3 Nonlinearityin the learning rule
B.4.9.4 Competition
B.4.9.5 Soft competition
B.4.9.6 Untrained neurons
B.4.9.7 Large competitive nets: further aspects
B.5 Continuous attractor networks
B.5.1 Introduction
B.5.2 The generic model of a continuous attractor network
B.5.3 Learning the synaptic strengths between the neurons that implement a continuous attractor network
B.5.4 The capacity of a continuous attractor network: multiple charts and packets
B.5.5 Continuous attractor models: path integration
B.5.6 Stabilization of the activity packet within the continuous attractor network when the agent is stationary
B.5.7 Continuous attractor networks in two or more dimensions
B.5.8 Mixed continuous and discrete attractor networks
B.6 Network dynamics: the integrate-and-fire approach
B.6.1 From discrete to continuous time
B.6.2 Continuous dynamics with discontinuities
B.6.3 An integrate-and-fire implementation
B.6.4 The speed of processing of one-layer attractor networks with integrate-and-fire neurons
B.6.5 The speed of processing of a four-layer hierarchical network with integrate-and-fire attractor dynamics in each layer
B.6.6 Spike response model
B.7 Network dynamics: introduction to the mean-field approach
B.8 Mean-field based neurodynamics
B.8.1 Population activity
B.8.2 The mean-field approach used in a model of decision-making
B.8.3 The model parameters used in the mean-field analyses of decision-making
B.8.4 A basic computational module based on biased competition
B.8.5 Multimodular neurodynamical architectures
B.9 Interacting attractor networks
B.10 Sequence memory implemented by adaptation in an attractor network
B.11 Perceptrons and one-layer error correction networks
B.11.1 Architecture and general description
B.11.2 Generic algorithm for a one-layer error correction network
B.11.3 Capability and limitations of single-layer error-correcting networks
B.11.3.1 Output neurons with continuous values, and random patterns
B.11.3.2 Output neurons with binary threshold activation functions
B.11.3.3 Gradient descent for neurons with continuousvaluedoutputs
B.11.4 Properties
B.11.4.1 Generalization
B.11.4.2 Graceful degradation or fault tolerance
B.11.4.3 Prototype extraction, extraction of central tendency, and noise reduction
B.11.4.4 Speed
B.11.4.5 Non-local learning rule
B.11.4.6 Interference
B.11.4.7 Expansion recoding
B.11.4.8 Utility of single-layer error-correcting networks in information processing by the brain
B.12 Multilayer perceptrons: backpropagation of error networks
B.12.1 Introduction
B.12.2 Architecture and algorithm
B.12.3 Properties of multilayer networks trained by error backpropagation
B.12.3.1 Arbitrary mappings
B.12.3.2 Fast operation
B.12.3.3 Learning speed
B.12.3.4 Number of hidden neurons and generalization
B.13 Deep learning using stochastic gradient descent
B.14 Deep convolutional networks
B.15 Contrastive Hebbian learning: the Boltzmann machine
B.16 Deep Belief Networks
B.17 Reinforcement learning
B.17.1 Associative reward-penalty algorithm of Barto and Sutton
B.17.1.1 Architecture
B.17.1.2 Operation
B.17.2 Reward prediction error or delta rule learning, and classical conditioning
B.17.3 Temporal Difference (TD) learning
B.18 Learning in the neocortex
B.19 Forgetting in cortical associative neural networks, and memory reconsolidation
B.20 Genes and self-organization build neural networks in the cortex
B.20.1 Introduction
B.20.2 Hypotheses about the genes that could build different types of neuronal network in different brain areas
Specification of neuron classes, and of connection rules between classes
Specification of synaptic learning rules
Specification of the operation of neurons
B.20.3 Genetic selection of neuronal network parameters to produce different network architectures with different functions
B.20.4 Simulation of the evolution of neural networks using a genetic algorithm
B.20.4.1 The neural networks
B.20.4.2 The specification of the genes
B.20.4.3 The genetic algorithm, and general procedure
B.20.4.4 Pattern association networks
B.20.4.5 Autoassociative networks
B.20.4.6 Competitive networks
B.20.5 Evaluation of the gene-based evolution of single-layer networks
B.20.6 The gene-based evolution of multi-layer cortical systems
B.20.7 Summary
B.21 Highlights
Appendix 3: Neuronal encoding, and information theory
C.1 Information theory and its use in the analysis of formal models
C.1.1 The information conveyed by definite statements
C.1.2 The information conveyed by probabilistic statements
C.1.3 Information sources, information channels, and information measures
C.1.4 The information carried by a neuronal response and its averages
C.1.4.1 A numerical example
C.1.5 The information conveyed by continuous variables
C.1.5.1 Example: the information retrieved from an autoassociative memory
C.2 Estimating the information carried by neuronal responses
C.2.1 The limited sampling problem
C.2.1.1 Smoothing or binning neuronal response data
C.2.1.2 The effects of limited sampling
C.2.2 Correction procedures for limited sampling
C.2.3 The information from multiple cells: decoding procedures
C.2.3.1 Decoding algorithms
C.2.4 Information in the correlations between the spikes of different cells: a decoding approach
C.2.5 Information in the correlations between the spikes of different cells: a second derivative approach
C.2.5.1 The correlations in the neuronal response variability from the average to each stimulus (sometimes called "noise" correlations) ?:
C.2.5.2 The correlations in the mean responses of the neurons across the set of stimuli (sometimes called "signal" correlations) ?:
C.2.5.3 Information in the crosscorrelationsin short time periods
C.3 Neuronal encoding: results obtained from applying information-theoretic analyses
C.3.1 The sparseness of the distributed encoding used by the brain
C.3.1.1 Single neuron sparseness as
C.3.1.2 Population sparseness ap
C.3.1.3 Comparisons of sparseness between areas: the hippocampus, insula, orbitofrontal cortex, and amygdala
C.3.2 The information from single neurons
C.3.3 The information from single neurons: temporal codes versus rate codes within the spike train of a single neuron
C.3.4 The information from single neurons: the speed of information transfer
C.3.5 The information from multiple cells: independent information versus redundancy across cells
C.3.6 Should one neuron be as discriminative as the whole organism, in object encoding systems?
C.3.7 The information from multiple cells: the effects of cross-correlations between cells
C.3.8 Conclusions on cortical neuronal encoding
C.4 Information theory terms - a short glossary
C.5 Highlights
Appendix 4: Simulation software for neuronal networks, and information analysis of neuronal encoding
D.1 Introduction
D.2 Autoassociation or attractor networks
D.2.1 Running the simulation
D.2.2 Exercises
D.3 Pattern association networks
D.3.1 Running the simulation
D.3.2 Exercises
D.4 Competitive networks and Self-Organizing Maps
D.4.1 Running the simulation
D.4.2 Exercises
D.5 Further developments
D.6 Matlab code for a tutorial version of VisNet
D.7 Matlab code for information analysis of neuronal encoding
D.8 Matlab code to illustrate the use of spatial view cells in navigation
D.9 The Automated Anatomical Labelling Atlas 3, AAL3
D.10 The extended Human Connectome Project extended atlas, HCPex
D.11 Highlights
Bibliography
Index

Systemvoraussetzungen

Als PDF speichern Als Link merken