Mail exchange regarding a paper by Usher and Donnelly. My comments on his answer.

I sent this mail (I lost the actual message):

Subject: Re: Your 'synchrony' paper Regarding your paper in Nature (Usher & Donnelly (9 July 1998),

I don't follow the logic of the conclusion about binding _inside_ the brain.. As far as I can see, it assumes that:

If humans are sensitive to X in the visual input, than X is used inside the brain.

[ And X in this case is simultanousity of the elements of a percpet ]

This looks to me totally invalid assumption, which is contrary to the facts (e.g. set X to size of objects, frequency of EM waves etc.). Can you explain this?

Yehouda Harpaz

This is his answer:

From: M.Usher_ukc.ac.uk
Date: Thu, 30 Jul 1998 14:11:45 +0100
To: yeh_harlequin.co.uk
Cc: M.Usher_ukc.ac.uk
Subject: Re: Your 'synchrony' paper Regarding  your paper in Nature (Usher & Donnelly (9 July 1998)

I don't follow the logic of the conclusion about binding _inside_ the brain.. As far as I can see, it assumes that: If humans are sensitive to X in the visual input, than X is used inside the brain. [ And X in this case is simultanousity of the elements of a percpet ] This looks to me totally invalid assumption, which is contrary to the facts (e.g. set X to size of objects, frequency of EM waves etc.). Can you explain this? Yehouda Harpaz

I think that you need to consider our statement about binding _inside_ the_brain, within its context which you eliminated from your nice explanation above. However I think that, first, even according to your presentation, there is a sense in which:

If humans are sensitive to X in the visual input, than X is used inside the brain.

This follows from the additional implicit assumtion that human perfornance is mediated by things which happen inside the brain. So for example, following your example, since the brain is not sensitive [meaning doesnt descriminate in its response] to frequencies of EM radio-waves, neither does human performance, and since it descriminates in its responses to the difference between red-blue visible light, so do human subjects.

What I understand that you imply is that, the actual brain mechanism responsible for performance, might be sensitive to the feature in the out_side world (in our case synchrony) only in an implicit, but not explicit way. But this is not the case for synchrony, since brain activity tends to lock on stimulus modulations (If you measure EEG when people watch a flickering display you find a peak in the spectrum at the frequency of the flicker).

The context in which our paper is stated is essential for the claims. Surely, only by looking inside the brain can one really know that binding is mediated by neural synchrony. However people looked inside the brain and reported those synchonised activities. The question was then, whether it could be relavant for the processes of grouping. Since visual modulation do triger modulations in responses of neurons, it was reasoned that one should be able to detect the effect of such manipulations in tests of grouping with human subjects. Its apriori possible that the performance is mediated by some other mechanism that depends on the temporal manipulations only implicitly. But then one needs to look at the details and compare the support for the alternative mechanisms. Most of our paper was dedicated to rule out such a mechanism (local motion computation) and we know of no other one (except grouping) that could mediate performance in this tasks.

Does this help?

-Marius