The Binding Problem-continued

Well, in my last post I kind of left my audience hanging in the fact that I discussed the binding problem, but didn’t give any proposed solution. In this post I want to discuss and speculate on a possible answer to the question of how vision is bound together.

Possibly, the most well-known solution is Treisman’s feature-integration theory, which is an attention-based theory of the human visual system. In a nutshell, Treisman has proposed that when you attend to an object, the fact that you are attending to it necessarily integrates all the salient features of the object together aka the features are bound together. Furthermore, he postulates “feature maps” in the parietal lobes of the brain are used to select the features being bound together for any particular object.

His theory can be tested and has been tested in the following way:

Two white digits are briefly presented in the center of a computer screen, one of which is physically large than the other; the subjects’ task is to report the larger of the two digits, a task that requires attention to be directed at the center of the screen. Simultaneously with the digits, two colored letters are briefly presented in the periphery, one of which is always an F or X accompanied by a distractor letter(such as an O). Thus, after reporting on the digits, subjects are asked which of the two target letters occurred(F or X) and, most importantly the color in which that target was presented. If attention is required for binding, one might expect to observe “illusory conjunctions” in this paradigm such that subjects miscombine the features making up the two peripheral letters. And, in fact, that is just what is observed-when a red O and a yellow X are presented, for example, subjects often report seeing a red X more often than would be predicted by chance.(Hunt & Ellis, 2004)

Furthermore, empirical support for Treisman’s theory has been found in patients who have sustained damage to the parietal lobes.

When presented with multiple objects, a patient [who had sustained bi-lateral damage to the parietal lobes] could only report the individual features making up various objects; he was unable to correctly report which features belonged to the same object!(Hunt & Ellis, 2004)

However, I’d like to point out that while there is a lot of empirical support for Treisman’s theory and various other cognitive/neural explanations, there is still an explanatory gap in the following way: as far as I am aware, no theory of vision that attempts to account for the binding problem adequately gives an evolutionary explanation. This is a problem because it seems logical for reasons of parsimony to assume that at some point in our evolutionary history salient features weren’t bound together, so in order to give a satisfactory answer to the binding problem, one must propose some sort of evolutionary pressure explaining how and why they got bound together. I will not go into details in this post, but I speculate that one can get around this “why” problem if one has a wider conceptual framework to substantiate “brain bound” theories such as Treisman’s. Without better conceptual frameworks, these neural theories of perception will necessarily have limited explanatory power, despite being supported by empirical evidence.

References:
Hunt, R., & Ellis, H. (2004). Fundamentals of Cognitive Psychology (Seven ed.): McGraw-Hill Higher Education.

The Binding Problem of Perception

Imagine looking at an apple. If asked to describe it you would probably begin describing it’s various features: its color, brightness, hue, shape, size, texture, spatial location, etc. Now suppose if someone asked you where are all these features located, you would probably give them a funny look. Isn’t it obvious? Out there! In the apple! They are the apple! That is what an apple is: a conglomeration of various features integrated into one continuous percept. This seems like a perfectly sensible explanation, but if this obstinate person continues his line of questioning and asks you where this apple-percept is located in the brain, you might have to crack open a textbook and get back to him, because now the answer is not so simple.

After studiously pouring over the latest research on visual processing, you are finally ready to give the questioner an answer: no where. Simply put, the various features that make up an “apple” are represented in a highly complicated manner across a dazzlingly diverse array of brain tissue. For the sake of simplicity, brain researcher’s often distinguish between two primary information pathways that sensory data takes: the what and where streams. These two streams form the basis of an exemplary conceptual framework for how the brain processes various features of the objects around us to form a more-or-less continuous percept of objects such as apples. It is these continuous percepts that allows us to manipulate and verbally describe them accurately.

So, if all these apple-features are neurally processed in a separated fashion, why do they appear to be bound together? One obvious answer is they are bound together for the sake of convenience for the perceiver, otherwise how could a person act in a meaningful way? In order to pick up an apple to eat it, a subject must have a more-or-less continuous perception of all the various apple-features integrated into a single spatio-temporal location. This seems to make sense, but who or what are these high-order representations being convenient for? Why would neurons care if things are bound together or not? Surely, from an evolutionary perspective, where efficiency and survival are involved, doing all this extra processing to bind all these neurally separate features together for some subject seems bizarre. Who is this perceiving subject and why would the brain decide to stop the important business of surviving and begin integrating all these salient features into a continuous perception? There needs to be a perceiver for there to be a perception and in order for there to be a perceiver, there needs to be some sort of integrated perception. It seems as if we are at a chicken-and-egg impasse, thus making the phenomenon of unity perception a problem for students of the mind.

There have been many proposed solutions to the binding problem over the years, but they are beyond the scope of this post, in which I only wanted to outline the problem. Whether or not I will attempt to discuss any of these solutions is yet to be determined as of now, but I wouldn’t be surprised if it was the focus of a forthcoming post. Sorry to leave you hanging!