Readings:
Marr, D. (1982). "The philosophy of the approach"
Kellman, P.J. & Arterberry, M. (1998). The cradle of knowledge: perceptual development in infancy
Vision is a perception of utmost importance in our everyday life. It is the primary channel through which we can learn about the physical properties of the objects around us and form our mental representations of them in our mind.
However, the study of visual perception is actually much more than simply studying about visual sensation or the biological basis of it. Of course, for our better understanding of how vision is possible, we will need the knowledge of how waves (photons) hit our photoreceptors in the retina and are transduced into electrical signals and sent to the visual cortices through the LGN and other neural pathways. It is also necessary for us to investigate at a deeper level of how those signals are being processed in the visual cortices, like what neurons are firing in what patterns in response to different forms of visual stimuli (e.g. edge detectors). Nevertheless, the sensational aspect and the detailed biological mechanisms behind vision are not actually what makes perception such a difficult subject to study.
Here, a crucial distinction between sensation and perception should be made. Basically, the process of how forms of energy (e.g. waves) in the outside world are received and transduced into the electrical signals by our sensational organs is part of the sensation, instead of perception mechanisms. But almost everything after that, from our depth perception to the recognition of objects and our ability to form mental representations of them are all parts of our perceptual system.
Then, the significance of understanding perception as information-processing at a more abstract level than detailed biological configurations should be stressed. This is what I really want to emphasize in this short essay, which is recognizing how perception is essentially an information-processing mechanism that could be potentially implemented in various kinds of hardware and thus having multiple realizability. When we are discussing about perception, instead of talking about the electrical signals in our brains, we are actually examining it at a more abstract level, involving how lines and shapes are perceived, how the 2-D image is transformed into a 3-D construction of the world, and how we filter out irrelevant information received from the sensational part and actively process only the task-relevant information. Absolutely, all of those processes mentioned above have some physical foundations, which means that there must be certain neuronal activities serving as the basis of this information-processing process in our brain. However, neurons are not the only way that this process could be achieved. Silicon chips might be as good as neurons, maybe even better.
In principle, what we are interested in is the computational/algorithm level, instead of the ecological or the implementational level (see Kellman’s article). And according to Marr, this level could not simply be inferred from the neural activities we discover in the brain. Personally, I do not completely agree with Marr’s viewpoint, but I recognize the distinction he has made between the two levels of analysis. I believe studying at the neural level should still be a possible approach and can at least provide us with a few insights regarding what kind of inputs and information are being processed on the computational level.