Monday Pets - Back to Basics: Visual Cognition (Here's one for the cat people)

Today for Monday Pets, we're going to go old school and talk about vision.

Vision is arguably our most (intentionally) utilized sensory system, so its pretty important to figure out how it works. And it's what David Hubel and Torsten Wiesel set out to investigate starting in the late 1950s. Ultimately, their work would get them a Nobel Prize in Physiology or Medicine, in 1981.

Basically, they took a bunch of cats, anesthetized them, and showed them patterns of light on a screen. Meanwhile, some microelectrodes were placed in various precisely determined spots in the cat's visual cortex, and they recorded the electrical activity.

Evil cats had it coming to them, I swear. (But seriously, animal research is important, and should only be used when there are no other reasonable alternatives.)

One question you might ask is: How do you know where the individual flower petals end, and the next one begins? How do you know where the flower ends and the bee begins? This is the $64k question, isn't it? Get too close to a flower and you might get stung. Was the beauty of the flower, however fleeting, worth the suffering imposed by the sting? And would a rose, by any other name, smell as sweet?

But I digress. We are here to talk about SCIENCE, and so we shall.

More generally, you might ask how you can detect any edge in the visual scene. As you might imagine, its a non-trivial question - the first step to any sort of object recognition is detecting where the object begins and ends within the scene. And without object recognition...well, life would suck. Big time.

By the time Hubel and Wiesel began their work, it was already known that visual cortex was organized retinotopically, meaning that each retinal cell projected to its own corresponding neuron in visual cortex. So if a spot of light is projected onto the retina, using single-unit recording with a microelectrode it is possible to locate the cortical neuron that responds to stimulation from that particular location in the visual field.

It was further known that that visual receptive fields were separated into excitatory "on" fields, and inhibitory "off" fields, and these fields were organized concentrically, with an on-center surrounded by an off-periphery, or vice-versa. If the "on" areas were stimulated with light, those neurons would fire. If the "off" areas were exposed to light, the firing of those neurons would be suppressed.

So you take a bunch of center-surround cells - let's say that these are ones with on-centers and off-surrounds - and you line them up. Now, if you project a bar of light into the retina that only intersects with one of the center-surround cells, that individual cell might fire, but that doesn't give you any edge information. But if that bar of light was rotated such that the retinal input coincides with that entire row of center-surround cells, then you're stimulating the on-centers from a whole row of cells. And since those cells are retinotopically organized, you know that the individual "dots" of light that stimulate each individual cell are actually organized in a line.

Ultimately, Hubel and Wiesel discovered that certain cells or cell columns were sensitive to lines (i.e. edges), but only at certain orientations. They discovered that certain cells were sensitive to lines at certain orientations, but only if they were moving across the visual field in a certain directions.

Sometimes a video can say more than any number of words or pictures.

It's long, but totally worth watching. Watch as the light moves across certain parts of the visual field, and listen. The clicking sound you hear is an audio representation of the electrical firing of neurons in the cat's visual cortex (though like any youtube video, the audio and video, at times, are slightly out of sync). Notice that certain cells only fire when the on-center is stimulated, but if you stimulate both the on-center and the off-surround, they cancel eachother out. Notice how they map out the orientation of various edge detectors, and certain cells sensitive to the motion of edges across the visual field. This is actual film footage from the laboratory of Hubel and Wiesel.

If you want to win a Nobel Prize, this is how. Do research that can be communicated so clearly with a video that no words are necessary.

Hubel DH, & Wiesel TN (2009). Republication of The Journal of Physiology (1959) 148, 574-591: Receptive fields of single neurones in the cat's striate cortex. 1959. The Journal of physiology, 587 (Pt 12), 2721-32 PMID: 19525558