G: Receptive Fields of Cortex

Some neurons responded to an oriented bar that could be positioned at various locations across an extended zone ...

... and responded best when it was moving!

How might the subjective state be known? Another opinion

1: Have neurons of the brain that respond selectively to a given ratio of cone activation. 2: If this is true there should be a color palate somewhere in the brain

How might the subjective state be known?

1: Have subjective state - one's conscious awareness of the color -- be determined directly by the ratio of the primary colors. 2: One possibility requires parallel channels that carry the primary colors of red, green and blue.

simple cells.

1: Note that most of the neurons in V1 will respond only when the stimulus lies at a particular location on the retina. 2: Also note that most of the neurons are "binocular", i.e., they respond to stimulation at corresponding locations from either eye.

Moving stripes activate a particular sector of the pinwheel column.

1: Pinwheels are shown using a new method of recording brain activity -- optical imaging. 2: Reflectance of the tissue changes when the cells are more active. This change is shown as pseudocolor in the activity map.

DAVID HUBEL TORSTEN WIESEL found

1: These results changes the view of what the cortex was doing. 2: It wasn't just providing a "screen" for projecting the image. Rather, it was beginning the process of identifying the elements that define the image.

v1 - The blobs contain lots of ______, so one can use an ____ to mark their location in primary visual cortex.

1: cytochrome oxidase 2:enzymatic reaction

In the adjacent cortex, known as V2, the cytochrome oxidase marker

1: shows as stripes. 2: The neurons in these stripes receive color information from the blobs.

ocular dominance columns, because each strip had input from only one eye.

1:After Hubel & Weisel discovered that columns were organized into hypercolumns, others found that the neurons in layer 4 of V1 received the input from lateral geniculate nucleus. 2: They were monocular and had center/surround receptive fields, not the elongated fields that register lines and edges. 3: Others also found that the input from a given eye connected to adjacent hypercolumns, and the hypercolumns serving a given eye formed a strip across V1.

TOPOGRAPHIC MAPPING OF THE IMAGE

A similar method was used autoradiography to label the neurons that were activated when the animals looked at the lines of the target. This work confirmed a long-held belief that the connections from retina to cortex provided for topographic mapping of the image

These rows of hypercolumns were called ocular dominance columns

And Hubel & Wiesel proposed a model wherein the brick-like hypercolumns that serve each zone of the retina are positioned like paving bricks across the surface of the cortex.

columnar organization

But when they penetrated vertically, the neurons which they encountered responded to the same orientation.

There are patients with damage to V4 who have lost

Color perception.

single opponent response

For a given color, there are cells where the center is excitatory and others where the surround is excitatory.

hypercolumn

Hubel & Wiesel later reported that the columns were organized in a systematic pattern, with the columns that serve a particular location of the retina being adjacent to one another.

Some of H&Ws early results proved to be wrong.

Later research indicated that the neurons of the fourth layer, which receive the input from LGN, do respond to spots of light (as do the retina and LGN neurons). 2: Further, though most of the other neurons provide a binocular response, i.e. they respond to an oriented bar that is presented to either or both eyes, the layer 4 neurons are predominantly monocular in their response.

Recordings from neurons in V4 do not show selective responses to various subtle hues.

Meaning No color palette!

Retina goes to ... Goes to ...

Retina ---lgn ---- cortex

complex cells

Some neurons responded to an oriented bar that could be positioned at various locations across an extended zone ... ... and responded best when it was moving 2: It is generally thought that complex cells receive their input from simple cells which serve adjacent regions of the retina.

tuning curve

The degree of activation depends on the orientation of the stimulus bar. If we designate the orientation that produces maximal activity as 0 degrees, then one will see progressively less activity as one rotates the stimulus away from that orientation. The plot of the relative activity is called the cell's

This illustration of cortical magnification can be confusing.

The dot map on the left shows that the image projects on the retina without distortions of size. The dot sizes on the right indicate how much cortical tissue (thus how many neurons) are devoted to each region of the retina, and thus to each region of visual space. With more neurons devoted to foveal images, one sees greater detail, i.e., there is higher acuity. 2: However, one's perception of the foveal image is not "magnified" so that it appears larger in size. Somehow size remains scaled properly across the full image - it is only the degree of detail that gets magnified.

And Hubel & Wiesel proposed a model wherein the brick-like hypercolumns that serve each zone of the retina are positioned like paving bricks across the surface of the cortex.

These rows of hypercolumns were called ocular dominance columns

DAVID HUBEL TORSTEN WIESEL were studying single-unit responses in primary and secondary cortex of cats, and later in monkeys.

They found that most of the cells in visual cortex do not have much spontaneous activity, and unlike the ganglion cells of the retina and cells in the LGN, the cortical cells do not respond vigorously to spots of light.

H&W also reported the startling result that the neurons of kittens (who had not yet opened their eyes) showed orientation selectivity.

This meant that the response properties of the neurons was determined -- in large part -- by innate development rather than learning. If, therefore, the response properties were determined by anatomical connections, then the anatomy of the newborn kitten was already specified in greater detail and with a precision which had not been previously imagined.

But new data suggests that a palette might exist in

V2.

DAVID HUBEL TORSTEN WIESEL

While one might speculate that the brain can use the output from ganglion receptive fields to identify edges and lines, there was no evidence in this regard until the work of Hubel and Wiesel, which began in the 1960s.

The rate of firing of the simple cells is a function of contrast. In other words,

a dark bar against a white background will fire more rapidly than will a gray bar against a dark gray background.

Similarly, a diagonal set of receptive fields converges upon an adjacent column, whose neurons thus respond to a bar which has

a diagonal orientation.

Some researchers and theorists have argued that orientation selective columns are not lined up, as suggested by the hypercolumn model of Hubel & Wiesel, but form

a ring around the blobs. 2: These modules would tile the surface of the cortex, presumably being organized in rows to provide the ocular dominance columns. 3: Recent work supports this view, and have renamed the modules. They are now called pinwheels

The colors used for blobs and stripes are completely

arbitrary. They do not specify the what colors the neurons are registering.

By the way -- as was described earlier, each hypercolumn has a

blob that contains the color sensitive cells for a given retinal zone. The blob is located in the center of each hypercolumn in layers 2 & 3.

These color responses are sent to primary visual cortex, mainly to a pool of neurons known as a

blob.

The subsequent work showed that B/W ganglion cells have

center/surround design.

We previously described the discovery that retinal ganglion cells have ______ (as well as cells that have black/white opponent responses.

color-opponent responses

As described previously, Hubel and Wiesel initially reported that neurons responded best to a bar that had a particular orientation, and there was

columnar organization of neurons within primary visual cortex. 2: The column would respond to a particular location on the retina, but there was initially no additional information about how the columns were organized in relation to one another.

Hubel & Wiesel hypothesized that the response of cortical cells was provided by

convergence from center/surround fields.

The neurons in the blob appear to combine inputs from overlapping single-opponent fields, for they show

double opponent responses

Rather, secondary cortex responded best to edges or to elongated bars ...

either a dark bar on a white background... 2: or a white bar on a dark background.

For a given color, there are cells where the center is .... and others where the surround is .....

excitatory and others where the surround is excitatory.

Zeki

has found that V4 is very important in color vision.

While one might speculate that the brain can use the output from ganglion receptive fields to

identify edges and lines, there was no evidence in this regard until the work of Hubel and Wiesel, which began in the 1960s.

The optic nerve synapses in the

lateral geniculate nucleus -- abbreviated LGN.

The result is that each half of the visual field is delivered to the

lateral geniculate nucleus and cortex on the opposite side of the brain.

For the right eye, the right half of the visual field goes to the, and the left half goes to the

nasal hemiretina 2:temporal hemiretina.

Extracellular recordings were showing that each eye sends information into

neighboring hypercolumns.

The LGN consists of six layers, each receiving input from ...Thus the input is said to be

one eye alone. monocular.

To better explain ocular dominance columns, we need to go back and discuss some basic anatomy about how the retina connects into the brain. Path to v1 =

optic nerve ---->lateral geniculate nucleus of the thalamus ----> v1

The amount of activation, reflected in rate of firing, is partly determined by the 1:, its 2 to the central axis, its...., and its length. So one cannot base one's judgment of orientation only on how rapidly a single neuron is firing.

orientation of the bar, 2: position relative 3: contrast 4: length

The response of the cortical neuron was described as being

orientation selective because the neuron was relatively unresponsive to bars that were tilted at other angles.

Note that most of the neurons in V1 will respond only when the stimulus lies at a

particular location on the retina.

In Hubel & Wiesel's earliest study, they penetrated the cortex at an angle, recording from many neurons, and observed what appeared to be a

random mapping of orientation selectivity.

We now know that the color responding ganglion cells also have center-surround design - a receptive field that responds in opposite ways to the opposite colors.

red vs. green cells are most abundant, blue vs. yellow cells are relatively rare (as are blue cones).

Theorists solve this problem by proposing that one's perception and judgments are based on the

relative firing rates among a pool of neurons (similar to trichromatic color theory).

Neurons in blobs and stripes show

similar responses.

For a given color, there are cells where the center is excitatory and others where the surround is excitatory.

single opponent response

V1 neurons differed in the width of their tuning curves

some being very wide and others being fairly narrow

Given what we now know about dual encoding by ON and OFF ganglion cells, and aware that surrounding colors affect what color is perceived, it appears that

the monochromatic color mixing work has not accounted for background (surround) color, or considered bi-directional encoding.

Optic nerve fibers from the temporal hemiretina do not cross at the optic chiasm ... ... but

the fibers from the nasal hemiretina do.

optic chiasm

the point in the brain where the visual field information from each eye "crosses over" to the appropriate side of the brain for processing

So for a column that is sensitive to vertical bars, the input to each cell in the column comes from

the vertical array of retinal receptive fields.

Further details about the discovery of ocular dominance columns required too much time to be put at the end of a lecture, so instead we discussed how the original hypercolumn concept was not quite right. Rather than putting the orientation columns into a row,

they lie in a ring - originally described as modules, and later described as pinwheels. But these would still tile V1, just as was proposed for hypercolumns.