Transcript ling411-11 - Rice University

Ling 411 – 11
Small-Scale Representation:
Cortical Columns
Perspective – What we know so far
Sources of information about the brain
 Aphasiology
• Research findings during a century-and-a-half
 Brain imaging
 Neuroanatomy
 Other research in neuroscience
• E.g., Mountcastle, Perceptual Neuroscience
(1998)
Perspective – What we know so far
 Large-scale representation
• Subsystems and their
•
 Locations
 Interconnections
The Wernicke Principle
 Cortical information processing
LARGE-SCALE REPRESENTATION
What we know so far – Subsystems I
 Phonology is separate from grammar and meaning
 Phonology has three components
•
•
•
Recognition (Wernicke’s area)
Production (Broca’s area)
Monitoring (Somatosensory mouth area)
•
•
Alternative pathways (cf. ‘phonics’ vs. ‘whole words’)
Angular gyrus
•
•
Different areas for different kinds of words
Different areas for the network of a single concept
•
In or near Broca’s area
 Writing likewise has three components
 Phonological-graphic correspondences
 Meaning is all over the cortex
 Grammar depends heavily on frontal lobe
LARGE-SCALE REPRESENTATION
What we know so far – Subsystems II
 Nouns and verbs are different
• In some ways (what ways?)
• How to explain?
 Written forms are connected to conceptual
information independently of phonological
forms
 Writing can be accessed from meaning even if
speech is impaired
 Conceptual information for nouns of different
categories may be in different locations
LARGE-SCALE REPRESENTATION
What we know so far – Subsystems III:
“Lexicon”
 The information pertaining to a single
lexical item is widely distributed
 That is, every lexical item is represented
by a large distributed functional web
• This web has subwebs for different kinds of
information
 Phonological (three subwebs)
 Multiple subwebs for different facets of the
meaning
LARGE-SCALE REPRESENTATION
What we know so far – Subsystems IV:
The Proximity Principle
 Neighboring areas for closely related
functions
• The closer the function the closer the proximity
 Intermediate areas for intermediate
functions
 Consequences
• Members of same category will be in same area
• Competitors will be neighbors in the same area
LARGE-SCALE REPRESENTATION
What we know so far – Subsystems V:
Locations of certain areas
 Locations of various kinds of “information”
• Primary
•
•
 Visual
 Auditory
 Tactile
 Motor
Phonological
 Recognition
 Production
 Monitoring
Etc.
PROCESSING
What we know so far – Processing
 Processing in the cortex is
• Distributed
• Parallel and serial
• Bidirectional
Next on the agenda:
I. Small-scale representation
Cortical Columns
II. Processing at the small scale
Operation of cortical columns
SMALL-SCALE REPRESENTATION
Findings that are now well established
 The brain is a network
 Composed, ultimately, of neurons
• Neurons are interconnected
•
 Axons (with branches)
 Dendrites (with branches)
Activity travels along neural pathways
Deductions from known facts
 Everything represented in the brain has
the form of a network
• (the “human information system”)
 Therefore a person’s linguistic and
conceptual system is a network
• (part of the information system)
 Every lexeme and every concept is a subnetwork
• Term: functional
web (Pulvermüller 2002)
Example: The concept DOG
 We know what a dog looks like
• Visual information, in occipital lobe
 We know what its bark sounds like
• Auditory information, in temporal lobe
 We know what its fur feels like
• Somatosensory information, in parietal lobe
 All of the above..
• constitute perceptual information
• are subwebs with many nodes each
• have to be interconnected into a larger web
• along with further web structure for
conceptual information
SMALL-SCALE REPRESENTATION
Findings not yet well established
 Cortical neurons are clustered in columns
• The column rather than the individual neuron as
•
the main operative unit
Each minicolumn acts as a unit
 Columns come in different sizes
• The smallest: minicolumn – 70-110 neurons
 When column becomes active all its neurons
are active
 Cortical columns as basic units of
• Information representation
• Processing
Quote from Mountcastle
“[T]he effective unit of operation…is not
the single neuron and its axon, but
bundles or groups of cells and their
axons with similar functional properties
and anatomical connections.”
Vernon Mountcastle,
Perceptual Neuroscience
(1998), p. 192
Three views of the gray matter
Different stains
show different
features
Layers of the Cortex
From top
to bottom,
about 3
mm
Layers of the cortex
 I – dendritic tufts of pyramidal neurons
• No cell bodies in this layer
 II, III – pyramidal neurons of these layers
project to other cortical areas
 IV – spiny stellate cells, receive activation
from thalamus and transmit it to other
neurons of same column
 V, VI – pyramidal neurons of these layers
project to subcortical areas
 Various kinds of inhibitory neurons are
distributed among the layers
Evidence for columns
 Microelectrode penetrations
 If perpendicular to cortical surface
• Neurons all of same response properties
 If not perpendicular
• Neurons of different response properties
Microelectrode penetrations in
the paw area of a cat’s cortex
Columns for orientation of lines (visual cortex)
Microelectrode
penetrations
K. Obermayer & G.G. Blasdell, 1993
The (Mini)Column
 Extends thru the six cortical layers
• Three to six mm in length
• The entire thickness of the cortex is accounted
for by the columns
 Roughly cylindrical in shape
 About 30–50 m in diameter
 If expanded by a factor of 100, the
dimensions would correspond to a tube with
diameter of 1/8 inch and length of one foot
Cortical Columns
(impressionistic
sketch)
A graphic model, not
an anatomical diagram
(There aren’t actually
any boundaries
between columns.)
From M. vanLandingham,
unpublished
Cortical column structure
 Minicolumn 30-50 microns diameter
 Recurrent axon collaterals of pyramidal
neurons activate other neurons in same
column
 Inhibitory neurons inhibit neurons of
neighboring columns
Columns and neurons
 At the small scale..
• Each column is a little network
 At a larger scale..
• Each column is a node of the cortical
network
 The cerebral cortex:
• Grey matter — columns of neurons
• White matter — inter-column connections
Minicolumns and Maxicolumns
 Minicolumn 30-50 microns diameter
 Maxicolumn – a contiguous bundle of
minicolums (typically around 100)
• 300-500 microns diameter
• Dimensions vary from one part of cortex to
•
another
In some areas at least, they are roughly
hexagonal
Cortical minicolumns: Quantities






Diameter of minicolumn: 30 microns
Neurons per minicolumn: 70-110 (avg. 75-80)
Minicolumns/mm2 of cortical surface: 1460
Minicolumns/cm2 of cortical surface: 146,000
Neurons under 1 sq mm of cortical surface: 110,000
Approximate number of minicolumns in Wernicke’s
area: 2,920,000 (at 20 sq cm for Wernicke’s area)
Cf. Mountcastle 1998: 96
Large-scale cortical anatomy
 The cortex in each hemisphere
• Appears to be a three-dimensional structure
• But it is actually very thin and very broad
 The grooves – sulci – are there because
the cortex is “crumpled” so it will fit inside
the skull
The cortical column as node
 Hypothesis: The nodes of a functional web
are cortical columns
 The properties of the cortical column are
approximately those described by Vernon
Mountcastle
• Mountcastle, Perceptual Neuroscience, 1998
 Additional properties of columns and
functional webs can be derived from
Mountcastle’s treatment together with
neurolinguistic findings
•
Method: “connecting the dots”
Topologically, the cortex of each
hemisphere (not including white
matter) is..
 Like a thick napkin, with
• Area of about 1300 square centimeters
 200 sq. in.
 2600 sq cm for whole cortex
• Thickness varying from 3 to 5 mm
• Subdivided into six layers
 Just looks 3-dimensional because it is
“crumpled” so that it will fit inside the skull
Topological essence of cortical structure
 The cortex is an array of nodes
• A two-dimensional structure of
interconnected nodes (columns)
 Third dimension for
• Internal structure of the nodes (columns)
• Cortico-cortical connections (white matter)
The cortex as a network of columns
 Each column represents a node
 The network is thus a large twodimensional array of nodes
 Nodes are connected to other nodes both
nearby and distant
• Connections to nearby nodes are either
•
excitatory or inhibitory
Connections to distant nodes are excitatory
 Via long (myelinated) axons of pyramidal
neurons
Simplified model of minicolumn I:
Activation of neurons in a column
Other
cortical
locations
Cell Types
II
III
Pyramidal
Spiny
Stellate
Thalamus
IV
Inhibitory
Connections to
neighboring
columns not
shown
V
VI
Subcortical
locations
Simplified model of minicolumn II:
Inhibition of competitors
Other
cortical
locations
Cell Types
II
III
Pyramidal
Spiny
Stellate
Thalamus
IV
Inhibitory
V
VI
Cells in
neighboring
columns
Another Quotation
“Every cellular study of the auditory
cortex in cat and monkey has
provided direct evidence for its
columnar organization.”
Vernon Mountcastle (1998:181)
Map of auditory areas in a cat’s cortex
A1
AAF – Anterior auditory field
A1 – Primary auditory field
PAF – Posterior auditory field
VPAF – Ventral posterior
auditory field
More quantities




Number of neurons in cortex: 27.4 billion
Number of minicolumns: 368 million
Neurons per minicolumn: average 75-80
Neurons beneath 1 mm2 of surface:
113,000
Mountcastle 96
Findings relating to columns
(Mountcastle, Perceptual Neuroscience, 1998)
 The column is the fundamental module of
perceptual systems
•
probably also of motor systems
•
Each column has a very specific local function
 Perceptual functions are very highly localized
 This columnar structure is found in all mammals
that have been investigated
 The theory is confirmed by detailed studies of
visual, auditory, and somatosensory perception in
living cat and monkey brains
Nodal interconnections
(known facts from neuroanatomy)
 Nodes (columns) are connected to
• Nearby nodes
• Distant nodes
 Connections to nearby nodes are either
excitatory or inhibitory
• Via horizontal axons (through gray matter)
 Connections to distant nodes are
excitatory only
• Via long (myelinated) axons of pyramidal
neurons
Local and distal connections
excitatory
inhibitory
Lateral inhibition
 Inhibitory connections
 Extend horizontally to other columns in
the vicinity
• These columns are natural competitors
 Enhances contrast
Inhibitory connections
Based on Mountcastle (1998)
 Columnar specificity is maintained by
pericolumnar inhibition (190)
• Activity in one column can suppress that in
its immediate neighbors (191)
 Inhibitory cells can also inhibit other inhibitory
cells (193)
 Inhibitory cells can connect to axons of other
cells (“axoaxonal connections”)
 Large basket cells send myelinated projections
as far as 1-2 mm horizontally (193)
Findings relating to columns
(Mountcastle, Perceptual Neuroscience, 1998)
 The column is the fundamental module of
perceptual systems
• probably also of motor systems
 This columnar structure is found in all
mammals that have been investigated
 The theory is confirmed by detailed
studies of visual, auditory, and
somatosensory perception in living cat
and monkey brains
Extrapolation to Language?
 Our knowledge of cortical columns comes
mostly from studies of perception in cats,
monkeys, and rats
 Such studies haven’t been done for
language
• Cats and monkeys don’t have language
• That kind of neurosurgical experiment isn’t
done on human beings
 Are they relevant to language anyway?
• Relevant if language uses similar cortical
•
structures
Relevant if linguistic functions are like
perceptual functions
Objection
 Cats and monkeys don’t have language
 Therefore language must have unique
properties of its structural representation
in the cortex
 Answer: Yes, language is different, but
• The differences are a consequence not of
•
different (local) structure but differences of
connectivity
The network does not have different kinds of
structure for different kinds of information
 Rather, different connectivities
Justifying extrapolation
 Hypothesis: Extrapolation of findings about cortical
columns can be extended to
•
•
humans
inguistic and conceptual structures
 Why? Cortical structure, viewed locally, is
• Uniform across mammalian species
• Uniform across different cortical regions
 Exceptions in primary visual and primary
auditory areas
 Different cortical regions have different
functions
• because of differences in connectivity
• not because of differences in structure
In particular..
 Cortical structure and function, locally,
are essentially the same in humans as in
cats and monkeys and rats
 Moreover, in humans,
• The regions that support language have the
same structure locally as other cortical
regions
Uniformity of cortical function
 Claims:
• Locally, all cortical processing is the same
• The apparent differences of function are
consequences of differences in larger-scale
connectivity
 Conclusion (if the claim is supported):
• Understanding language, even at higher levels,
is basically a perceptual process
Argument for local uniformity
of representation



Different types of cortical information
•
•
•
•
How are they different?
Two possibilities
2.
They differ in their structural form
They differ based on their connections
•
The “connectionist claim”
1.

Perceptual
Conceptual
Grammatical
Phonological
Claim: Possibility #2 is the correct one
Support for the connectionist claim
 Lines and nodes (i.e., columns) are
approximately the same all over
 Uniformity of cortical structure
•
•
•
Same kinds of columnar structure
Same kinds of neurons
Same kinds of connections
 Conclusion: Different areas have
different functions because of what
they are connected to
Perception and Language


Our knowledge of cortical columns comes
mostly from studies of perception in cats,
monkeys, and rats
Why haven’t such studies been done for
language?
1. That kind of neurosurgical experiment isn’t
done on human beings
2. Cats and monkeys don’t have language

Are they relevant to language anyway?
1. Relevant if language uses similar cortical
structures
2. Relevant if linguistic functions are like
perceptual functions
Relevance to Language
 These studies of perception are
relevant if
• Perceptual structure and functions are
•
basically the same across modalities
 Including associative areas (higherlevel)
Linguistic comprehension is basically a
perceptual process
Linguistic Information in the Cortex
 Problem: Linguistic information is usually
described symbolically
 In the symbolic mode of description,
different kinds of linguistic information
appear to have different kinds of structure
•
•
•
•
Phonology
Morphology
 Regular and irregular inflections
Syntax
Semantics
 Claim: If the information is viewed as
connectional instead of symbolic, it turns out
to have a high degree of uniformity
Uniformity of cortical structure






Six layers throughout, with similar structure
Columns throughout
Same neuron types everywhere – pyramidal most
frequent, spiny stellate in layer IV, etc.
Inhibitory and excitatory connections throughout
Same neurotransmitters everywhere
•
•
Excitatory: glutamate
Inhibitory: GABA
But: What about the different Brodmann areas?
1. The differences are relatively minor
2. They may be based on experience
Structural Uniformity?
A closer look
 Differences are found at lower levels
• Primary sensory areas have specialized
•
•
structures
These are called heterotypical areas
Properties of columns depend on afferent inflow
 More uniformity in higher-level areas
• “Homotypical” areas (i.e., same type)
• Relatively uniform structure
• Makes them flexible, adaptable
• Properties depend on intracortical processing
• Different homotypical areas differ not because
of their structures but because of their
connections
A heterotypical area: Visual motion perception
An area in the
posterior bank
of the superior
temporal sulcus
of a macaque
monkey (“V-5”)
Albright et al. 1984
Auditory areas in a cat’s cortex
(Heterotypical)
A1
AAF – Anterior auditory field
A1 – Primary auditory field
PAF – Posterior auditory field
VPAF – Ventral posterior
auditory field
Exceptions: Diversity in cortical function
 Perception vs. production
• Back brain vs. front brain
 Sharpness of contrast
• Phonology and morphology require sharp contrasts
• Conceptual categories have fuzzy definitions
 Left vs. right hemisphere
• RH may have..
 Larger minicolumns
 Less lateral inhibition
Uniformity in LH Associative Areas
 The associative areas are homotypical
 The structure that subserves language
understanding is the same as perceptual
structure
• Columns of similar structure
• With similar kinds of connections
 Claim: Understanding language is the
same process as perception
• To support this claim, must look more closely
•
at cortical function
Subclaim: Locally, all columns function alike
The uniformity of information?



Different types of cortical information
•
•
•
•
How are they different?
Two possibilities
2.
They differ in their form of representation
They differ based on their connections
•
The “connectivity claim”
1.

Perceptual
Conceptual
Grammatical
Phonological
Claim: Possibility #2 is the correct one
The uniformity of cortical function
 Claims:
• Locally, all cortical processing is the same
• The apparent differences of function are
consequences of differences in larger-scale
connectivity
 Conclusion (if the claim is supported):
• Understanding language, even at higher levels,
is basically a perceptual process
Testing the claim
 Claim:
• The apparent differences of function are
consequences of differences in larger-scale
connectivity
 To test, we need to understand cortical
function
 That means we have to understand the
function of the cortical column
Columns do not store symbols!
 They only
• Receive activation
• Maintain activation
• Inhibit competitors
• Transmit activation
 Important consequence:
• We have linguistic information represented
•
in the cortex without the use of symbols
It’s all in the connectivity
 Challenge:
• How?
end