orientation neurons

[gib65]

I'm reading a paper that talks about the anatomy of our visual system. It says that there are neurons in the occipital lobe corresponding to our perception of orientation of lines. That is, when we see a series of points in a continuous straight line, these cells become active, and moreover there is a range of cell "columns", each corresponding to a particular orientation of lines. My question is this: when each point on our retina sends information to the occipital lobe, do they each get their own full set of line-orientation neurons, or is there only one set of line-orientation neurons in the brain that each point sends its information to?

[ashennell]

My question is this: when each point on our retina sends information to the occipital lobe, do they each get their own full set of line-orientation neurons, or is there only one set of line-orientation neurons in the brain that each point sends its information to?

 

The projection from the retina via the thalamus to the cortex is topographic. So the organisation of visual information in the occipital lobe is retinotopic. So the short answer is that each point on the retina goes to different 'line detector' neurons. However, this is unlikely to be perfectly point-to-point.

 

In fact, there are many retinotopic maps in the primary visual areas - i.e. there are parallel projections that form multiple retinotopic maps. I fail to recall how many there are and to what extent they differ in resolution.

[gib65]

So the short answer is that each point on the retina goes to different 'line detector' neurons.

 

So each retinal point gets its own fully set of line-orientation neurons in the occipital lobe. Am I interpreting you correctly?

[ashennell]

So each retinal point gets its own fully set of line-orientation neurons in the occipital lobe. Am I interpreting you correctly?

 

roughly, yes. It's not a perfect mapping where every retinal neuron gets a perfect set of orientation cells just for itself. It is an approximation of this pattern. We call this organisation retinotopic. Neurons that are next to each other in the retina will project to adjacent parts of the visual cortex (with overlap).

 

note : to detect a line in the cortex, a cortical neuron must receive input from a set of adjacent retinal cells - a line of them at least. SO line detection 'per se' requires that there is some overlap in the projection. Do you understand?

 

I think I need I diagram.

[gib65]

No need. I understand. Thanks.

[gib65]

Now I've got a related question: What about the perception of curves? Are their neurons specializing in this? Do signals have to pass through the line detectors before being perceived as curves, or do we perceive curves independently from lines?

[ashennell]

I'm not too sure about curve detection.

 

I don't think that there are cells in the primary visual cortex that respond specifically to curves of different orientation or 'radius'. I think that curve detection is probably achived in the next layer up in the cortical hierarchy - the secondary visual areas. Probably through recognition of patterns of short line segments that are produced from the primary cortical area.

 

It may be worth noting that some neurons in the primary visual cortex actually respond to end-stopped lines. Lines of a certain orientation that stop within the cells receptive field. These may assist in curve detection as a curve can be approximated by a series of short lines.

 

I'm not sure if this is clear - but I must dash. I you want any more info please feel free to ask.