What is the binding problem? In neuroscience what we mean by binding is tying together the activity of neurons that may be in distant parts of the brain. We know that this is likely to be crucial because the brain represents things in a distributed fashion.
It may be the case that two properties (say colour and shape) belonging to a perceptualy unified object (e.g. the percept of a motorbike) are coded in different parts of the brain. This raises some important issues that neuroscientists are still puzzling over.
I have been trying to think of an analogy that clearly illustrates why binding is a problem, and how potential solutions may work. This is the best I have come up with....
Think of two black bags - one full of cards with a different colour painted on each. The other full of cards with a different shape painted on each, such as cars, people or animals etc (all drawn in black and white). If the idea of discrete neural assemblies is true, when the brain represents something it is as though one hand is placed inside each bag and a card selected. The left hand may select a red card from the colour bag, and the right may select a picture of a car from the picture bag. Remember though that we can't take the cards out of the bags, we can only peek in through the gap at the opening. So, having just one feature conjunction being represented is no problem - we can peek in and see red in one bag, and a car in the other. What must be out there in the real world is a red car. However, consider the situation where we are trying to represent multiple objects. Say, as well as a red car, we are also looking at a green bike. In that situation we must select a green card, and a red card from the colour bag, and the pictures of both a car and a bike from the form bag. But how do we know which colour goes with which picture? This question lies at the heart of the binding problem. How is it that we can correctly represent a red car and a green bike, instead of a green car and a red bike? The latter situation is what is known as an illusory conjunction. Pairing the wrong features and therefore generating an 'illusory' percept.
There are two alternative solutions to this currently considered plausible by brain researchers. 1) Attention acts as a gating mechanism and selects the features of a single object for processing at any one instant, or 2) the temporal pattern of activity forms another component of the neural code that is used to code relational information.
To illustrate the attentional hypothesis (1) lets go back to our analogy with bags and cards...Imagine that instead of being forced to keep our hands deep inside the bags we are allowed to select a single card from each bag and remove it. Then we can put the two cards down on the table in front of us and stick them together. This, of course, will make it very easy to see which colour goes with which picture, but comes with the rather large drawback that only a single perceptual object can be active at one time. Amongst psychologists this is known as the 'feature integration theory' of attention, proposed by Anne Treisman in the 80s.
The second possibility is newer (and more contentious). It has been proposed that synchronous oscillations are crucial for binding (von der Malsburg & Wolf Singer being the original proponents). Oscillatory patterns in neural firing have been reported for many years but the functional relevance of this is still a matter of debate, and many research labs are attempting to answer this question. If neurons A B and C all fire at the same rate their activity is difficult to distinguish. However, if A and B fire periodically at the same time points (in phase), whilst neuron C fires at different time points (e.g. in antiphase), this may be used as a type of binding information. Going back to the analogy, this is akin to somehow weaving a thread between a coloured card in one bag, and connecting it to a picture in the other bag. Many of these threads can be fastened at a single instant, so multiple objects can be represented simultaneously, and the thread can rapidly be tied and untied.
I hope this analogy is helpful in understanding what neural binding is all about!
No comments:
Post a Comment