BRAIN NETWORKING Flashcards
Striatum of Basal Ganglia
putamen and caudate nucleus
Basal ganglia contributes to:
– Action selection
– Reinforcement learning
Nearly all of the cerebral cortex projects to the striatum except for the primary visual cortex and primary auditory cortex
Increased striatal activity can disinhibit thalamus (via direct pathway)
Striatum inhibits GP internal segment, which removes inhibition of thalamus
Hyperdirect pathway
cortex to subthalamic nucleus
Direct pathway
striatum to GP internal segment
Indirect pathway
Striatum to GP external segment to subthalamic nucleus to GP internal segment
Cerebellum plays a role in more automatic execution during/after skill learning
Cerebellum involved in both motor and cognitive functions
Cerebellum receives copies of commands from motor and prefrontal cortex
Copies of commands called “efference copies”
Cerebellum may output predicted sensory consequence of movements?
Cerebellum may predict new state of body based on efference copy?
Hippocampus functions
– Episodic memory
– Spatial navigation
Parahippocampal areas
parahippocampal cortex,
perirhinal cortex, entorhinal cortex
“Six degrees of separation”
Idea that everyone can be connected in ≤ 6 steps
“The small world problem”
Regular network
Every node (dot above) connected to its nearest neighbors (nearby dots)
Random network
Increase disorder by reconnecting edges to random nodes until all edges are wired randomly
Connection is called “edge”
Small-world network
High clustering like regular graph, yet small characteristic path length (Global average of all distances) like random graph
Module
a subset of nodes with high within-module connectivity and low inter-module connectivity
Path length
minimum number of edges to go from one node to another
Node degree
Number of connections that link a node to the rest of the network
Clustering coefficient
Number of connections that exist between nearest neighbors of a node (as a proportion of the maximum number of possible connections)
Rich-club architecture
Type of small world network evident in the brain
Rich node
Node with a large number of connections, i.e., high-degree node (called network hub)
Rich club
Rich nodes that are well-connected with each other, forming a tight subgraph
Rich-club organization
Greater likelihood of high-degree nodes forming clubs than low-degree nodes
Anatomical connections
– Axon projects from one neuron to another
– Parallel projections of axons form white matter paths in the brain
How to measure anatomical connections
– Tracer studies (invasive): Tracer molecule injected into brain and travels along axons
– Diffusion MRI (non-invasive): Infer direction of white matter path based on water diffusion
Functional connections
– Correlated neural activity between different brain areas
– Functional connection may reflect direct or indirect anatomical path between brain areas
How to measure functional connections
– Statistical dependencies: correlation, coherence (correlation between oscillation frequencies)…
– Can measure functional connections based on spike rate, local field potentials, EEG, functional MRI (BOLD activity)
Diffusion MRI
Water diffuses more readily along connections between brain areas
Measure water diffusion in each brain voxel
Connect voxels based on preferred diffusion directions
Diffusion MRI techniques include DTI or diffusion tensor imaging
Functional MRI
Parcellate brain into regions of interest (ROIs)
Calculate correlation between BOLD activity in two ROIs
Repeat correlation calculation for all possible pairs of ROIs
Resting-state functional MRI
Subject scanned with no behavioral task
Functional connectivity is time-dependent and modulated by task context
Schizophrenia
characterized by delusions, hallucinations, disorganized speech, other symptoms that cause social or occupational dysfunction
position invariance
can identify an object no matter where it is
Sensor with big receptive field
How to operate effectively on multiple spatial scales?
– How to localize and identify small objects or parts of objects?
– How to identify big objects or scenes?
– How to identify objects wherever they are (position invariance)?
topographic maps
Multiple representations of the environment
– Representation of our environment built from small receptive fields
– Representation of our environment built from big receptive fields
– Representation of the environment built from different stimulus features, e.g., motion
– Orderly representation of sensory space
– Different neurons have receptive fields covering different parts of sensory space
– Neurons arranged such that nearby neurons represent nearby regions of sensory space and distant neurons represent distant regions of sensory space
How to build receptive fields of different sizes?
– Small receptive fields usually found at early stages of sensory pathways near sensory organs
– Bigger receptive fields can be built from small receptive fields
– If neurons with small, adjacent receptive fields all provide input to the same neuron, then summing these inputs will produce a bigger receptive field
Somatosensory RF
– area of body surface
– smallest RFs for finger tips
– largest RFs for thigh/calf
Mapped along dimensions of body surface
Visual RF
– area of visual space
– smallest RFs only a few minutes of arc (like dot on page)
– largest RFs tens of degrees (like entire page of book)
Mapped along (usually 2) dimensions of the space around you
receptive fields (RFs)
part of sensory world to which neuron responds
Olfactory receptive field
Mapped along dimension of carbon chain length of odorant
Numerical receptive field
space on an abstract scale
Found in some parietal and prefrontal neurons
Mapped along dimension of numerosity
big receptive fields
– to identify objects
– for position invariance
small receptive fields
– to identify detailed features of an object
– for high acuity
Fovea
part of retina with high spatial resolution
– Certain parts of sensory space may occupy disproportionately large part of map, e.g., fovea
– This allows greater sensitivity for those parts of sensory space
central part of visual field
Brain areas commonly defined by their representation of sensory space
– Sensory space is mapped multiple times in the brain
– E.g., there are multiple visual maps of the space around us
– Each distinct visual brain area (V1, V2, etc) contains a complete representation of half of visual space (called “hemifield”)
– I.e., visual area in left hemisphere predominantly represents right visual field (and vice versa)
Retinotopic map
– orderly representation of visual space (hemifield)
– called “retinotopic” as it reflects organization of retina
Hemifield = half of visual space
Tonotopic map
auditory brain areas
orderly representation of sound (tone) frequency
Somatotopic map
orderly representation of body surface
Efficient design to group neurons together that are highly interconnected
– Neurons processing nearby sensory space will interact more often (than with other neurons)
– Grouping these neurons together reduces wiring
connections within a map?
– Each neuron in a map is connected with a subset of the other neurons
– If the number of connections of each neuron is held constant, then the proportion of the map that each neuron connects with depends on the number of neurons in the map
Large number of neurons in fine-grained maps
small RFs
– Each neuron requires more and longer connections?
or
– Each neuron connects with fewer neurons?
Fewer neurons in coarse-grained map
big RFs
– Neurons representing distal parts of the map more readily connected
– Facilitates comparison/integration of information from different parts of the map
