The Ventral Stream

- the 'what'

In the two-streams hypothesis, the ventral stream is also referred to as the 'what pathway'. It's thought to play a key role in object recognition. It can be traced back to the P, or midget retinal ganglion cells of the retina, which receive input from relatively few cones in the photoreceptor layer located above the foveal region of the retina. It continues through the LGN and into V1, from where it radiates in the ventral-anterior direction; which means forward, along the bottom half of the brain. Its first stops are areas V2 and V4, before termination in a particular region of the temporal lobe called the inferotemporal (IT) cortex.

The IT cortex is made up of three gyri, all of which are present on the temporal lobe, which, incidentally, is a lobe unique to primates. The IT cortex is a long structure. Its posterior end is marked off by the preoccipital notch, and it winds all the way along the bottom and under the temporal lobe, traveling up its inward-facing surface before stopping at the inferior sulcus. The top of the IT cortex is made up of the middle temporal gyrus. Its middle part contains the inferior temporal gyrus, part of which curves under the lobe and is present on the internal surface. The last part of the IT cortex, the fusiform gyrus, is located on the internal surface of the temporal lobe.

The brain expanse the IT cortex is composed of can be broken down into multiple sections. From our perspective, as we travel from the back of the brain toward its front, the first area is the TEO, which demarks the most posterior part. The second is the TE, which is the anterior portion. Because more segmentation is always a good idea, the TE is divided into an anterior segment TEa, and a posterior one, TEp.

A huge amount of interest in the IT cortex was sparked by the discovery that individual neurons composing it responded not to elemental forms - bars and spots of light, which can build up into images, like lego pieces - but rather to whole objects, straightaway. These highly specialized neurons appear to be grouped in distinct regions. In monkeys, for example, a region of the IT cortex has been determined which responds to the faces of other monkeys. 97% of the neurons within this area demonstrated this face-specialization. A homologous (serving the same function) area was subsequently found in humans, in whom it is now termed the fusiform face area.

The IT isn't only concerned with faces. Different regions within it seem to respond to different categories of objects. In humans, for example, neurons that recognize buildings and landscapes are grouped together in parahippocampal place area. Experiments have shown that the further you test along the IT gyrus in the anterior direction, the images neurons respond to become more and more complex.

A cool fMRI experiment used the visual illusion below to suggest that the neurons of the IT cortex don't just respond to the form of an object, in the manner that V1 neurons respond to the form of a bar of light, or LGN neurons to the form of a spot of light. IT neurons go a step further - they respond to the perception of an object category. If you look at the picture below, sometimes you can see a vase, and sometimes a face. When you can see the face, neurons inside your fusiform face area are twinkling away like crazy. But when you start seeing the vase, they quiet down. If they were sensitive to the shape of the face only, they would never 'see' the difference.

The association of particular IT neurons to particular objects within an area of the IT specialized to an object category has been demonstrated in clever ways. For example, the presentation of the same object to a person will activate some specific area in the IT cortex. Transforming the visual conditions - like illumination, viewpoint angle, size, etc. - but not the object itself won't change which neurons respond within that area. Changing the object but not the object-category, however, will activate a different population of neurons; but not a different IT area. This is a phenomenon known as response invariance.

One thing that's worth keeping in mind is that much of the theory behind the two-streams hypothesis comes from lesion studies in humans. As a consequence, the studies have a small sample size - it's hard to find people with brain damage sustained in just the brain areas a researcher would like to investigate. in fact, the foundations of most of what I've outlined above were formed from tests done on just one patient, a woman known as 'DF', who had suffered carbon monoxide poisoning and had areas in her ventral stream irreversibly damaged as a result.

Considering her case gives one a good anecdotal idea of the roles of the two visual streams. For example, DF is unable to distinguish between two blocks of different sizes. She can never tell which one is the smaller or the larger, and fails completely when asked to indicate the approximate width of a block with her thumb and index finger. Yet, when simply directed to reach and grab for one of the blocks, she'll do it seamlessly; her hand will adapt the suitable size for picking it up. Her dorsal stream, in other words, is intact.

Similar tests have been done to test DF's ability to perceive orientation. Slits were made into a surface in multiple different directions, and she was asked both to verbally describe and physically indicate the directions of these slits. She couldn't do either; yet when told to place her hand into one of these slits, she performed as a normal subject would, twisting her hand in the right direction from the very start of the movement.

Monkeys with ventral stream damage suffer in what researchers think of as analogous ways. They too lose the ability to recognize objects. But they remain extremely adept at other visually challenging tasks, like catching flies.

The case of DF highlights a key recurring theme of the central nervous system - systems within it impinge on consciousness in varying degrees. The autonomic nervous system controls 'subconscious' systems used in breathing, the beating of your heart, or hormone release. It does all the work we're too stupid to be in charge of, and can do it because it isn't centralized and delegates its responsibilities to each of the millions of tiny components it's made up of. The somatic nervous system, on the other hand, is voluntary, and thus probably what you associate your concept of selfhood with. The two proposed streams of the visual system are a good example of this dichotomy falling apart. Their interrelation shows quite clearly that in large part, your body dictates what your self can be.

Damage in the Ventral Stream

Prosopagnosia: In Greek, 'prosopon' means 'face'. An agnosia, in medical terminology, is the inability to process sensory information and hence to recognize things. Prosopagnosia, or face blindness, is one of the more famous neurological disorders. People afflicted by are not able to recognize familiar faces; they can't even recognize themselves when prompted by a photograph. Prosopagnosics use other visual and non-visual cues, like gait, clothing, or the sound of a voice, to recognize the people they interact with.

The disorder is linked with the fusiform face area discussed above. It can be congenital or acquired.