Developing Intelligence

The infamous “binding problem” concerns how a coherent subjective experience of the world can emerge from the widely-distributed processing of individual object characteristics (for example, object identity and object spatial locations appear to be processed by independent neural systems). It is clear that binding requires focused attention (at least, according to “Feature Integration Theory”), but the specific mechanism by which attention binds remains elusive. Partly because of this, Baddeley has even revised the standard model of working memory to include an “episodic buffer” for binding.

A different, but still controversial solution to this problem concerns “multiplexed neural synchrony” – the idea is that neurons representing a specific object will fire synchronously yet out of phase with the neurons representing different objects. Those who disagree with this hypothesis typically suggest that multiplexed synchrony requires more temporal precision than possible in biological neural networks.

An alternative hypothesis is pursued in Colzato, Wouwe & Hommel’s recently accepted manuscript at Neuropsychologia. According to this framework, visuomotor binding is accomplished by the dopaminergic system. The authors used spontaneous eye blink rate to index central dopamine levels (which is an extremely surprising yet reliable index) in combination with repetition priming paradigm to test binding.


Colzato et al. asked 18 subjects to perform the following simple task. First, they will be cued to press a right- or left-hand button when a visual stimulus appears; this visual stimulus can vary in shape, location, and color, but these characteristics are irrelevant for the keypress (which I’ll refer to as R1). Three seconds later, the subject encounters a second visual stimulus which also varies in shape, location and color, which requires a left or right-hand keypress (R2) depending on the shape of the stimulus (all the other visual characteristics are irrelevant).

Subjects are typically slower to respond to the second stimulus if the shape-response binding is different between the first and second responses. In other words, they show a partial-repetition cost such that reaction time is slower on R2 if either the shape or the response (but not both) is different than that for R1. Similarly, R2 is slowed if the shape or the location (but not both) is different than on R1.

Why should this be the case? Colzato et al. argue that subjects are still binding together the stimulus features (in the case where shape or location differs) or the visuomotor features (in the case where response or shape differs), and that this interferes with their ability to accept a different visual or visuomotor binding on R2. This lingering binding results in a partial repetition cost.

The results showed that individual dopamine levels, as indexed by eye-blink rate, correlated with the amount of partial repetition cost, but only for the situations where the mapping between shape and response changed between R1 and R2. In other words, dopamine appears sensitive to visuomotor but not purely visual binding.

The authors suggest that dopamine relates only to binding in the visuomotor domain, and propose that choline relates to binding in the purely visual domain. This is a very strong hypothesis given that central dopamine may have multiple diverse effects on cortical processing, particularly those reliant on amodal regions of the prefrontal cortex, which are not necessarily involved in motor activity.

As you might expect, these results can be interpreted in a variety of other ways. For example, perhaps dopamine relates only to binding of task-relevant features, and has little to do with the motor system. Similarly, perhaps dopamine is involved in inhibition, and those with more dopamine were less able to inhibit the previous task (although this would run in the opposite direction of most inhibition accounts of dopamine function, given that this sample appeared to have relatively low blink rates compared to previous work).

One difficulty in interpreting these results is that the binding task could also be viewed as a task-switching paradigm. For example, subjects must go from a memory task (remember the cue) to a feature judgment task (respond to shape). This would be less troublesome if the authors reported whether color-location partial repetitions resulted in a cost, because that would control for changes in task – those features are irrelevant in both R1 and R2. A similar phenomenon has been discussed in the context of negative priming, where subjects are slower to respond to previously ignored features of an object – explained in terms of “selection feature mismatch.” The extent to which these phenomena overlap is an interesting area for future research.

Edit: I forgot to mention that the “feature mismatch” and dopaminergic accounts of binding are not necessarily in conflict with the multiplexed synchrony model mentioned above. Indeed, mismatching features may cause a problem precisely because of differing temporal coherence between features; similarly, dopamine is known to be important for prefrontal “up states” of activation, and could therefore also mediate the coherence or synchrony of firing in prefrontal cortex that supposedly accomplishes binding, according to the multiplexed synchrony hypothesis.

Related Posts:
Eyes, Window to the Soul – And Dopamine Levels?
The Argument for Multiplexed Synchrony
A Candidate Neural Mechanism for Cross-Frequency Phase Coupling
Non-Inhibitory Accounts of Negative Priming
The Development of Visual Binding