Binding Problem - The Practical Segregation Problem

The Practical Segregation Problem

In the case of visual perception, the brains of humans and other animals process different aspects of perception by separating information about those aspects and processing them in distinct regions of the brain. For example, Zeki and coworkers have shown that different areas in the visual cortex specialize in processing the different aspects of colour, motion, and shape. This type of modular coding yields a potential for ambiguity in many instances. Thus, when humans view a scene containing a blue square and a yellow circle, some neurons signal in response to blue, others signal in response to yellow, still others to a square shape and a circle shape. Here, the binding problem is the issue of how the brain represents the pairing of colour and shape, i.e. is the square blue or yellow?

Presumably somewhere further down the line of interneural signalling the output of the blue sensitive cells and the square sensitive cells are allowed to interact by converging in the inputs of further cells in a way that the output of the blue sensitive cells and the circle sensitive cells are not. Whether this is achieved by timing or some form of gating still appears to be wide open to debate. Relevant to this issue is the work of Pylyshyn, which suggests that incoming visual (and other) data get allocated "object labels" in some way, such that "blue" and "square" get tagged as "object number 1". In fact this labelling is so potent that if a blue square goes behind a screen and a yellow square appears at the other side the perception is that "object number 1 changed colour".

A popular hypothesis is that features are bound via synchronisation of the firing of different neurons in the cortex. Andreas K. Engel and his coworkers have found that two different neurons with a different receptive field produce divergent correlograms according to whether the stimuli were bound together or not. However, Thiele and Stoner found that perceptual binding of two moving patterns had no effect on synchronisation of the neurons responding to the two patterns. In the primary visual cortex, Dong et al. found that whether two neurons were responding to contours of the same shape or different shapes had no significant effect on neural synchrony.

A number of people, including Marcus have pointed out that phase of oscillation would only allow segregation of very few aspects of a visual image. At the very most ten phase states might be distinguishable, and even that would require stable oscillation rates over time. A much more plausible explanation would seem to be some sort of gating of allowed connection paths by short-term modulation of the weighting of synaptic inputs from "colour and shape cells" to downstream cells by signals from "object label cells".

In this form the binding problem is also an issue in memory. How do we remember the associations among different elements of an event? How does the brain create and maintain those associations? Both the hippocampus and prefrontal cortex seem to be important for memory binding. There is an implication that the behaviour of "object label" or perhaps "event label" cells, can be converted into some form of long-term reinforcement capacity.

In her feature integration theory, Anne Treisman suggested that binding between features is mediated by the features' links to a common location. Psychophysical demonstrations of binding failures under conditions of full attention provide support for the idea that binding is accomplished through common location tags.