Re: Images Vs. Symbols

From: HARNAD Stevan (
Date: Thu May 30 1996 - 21:36:52 BST

> From: "Spencer Klair-Louise" <>
> Date: Wed, 22 May 1996 11:29:31 GMT
> The issue in the imagery is not whether brain images exist;

You mean mental images; brain images are PET and MRI scan images.

> (we each
> have our own personal experience of imagery as testimony to this), but
> whether mental imagery can best be explained by manipulation of
> arbitrary symbols (as propositionalists such as Pylyshyn claim) or by
> analogue proccessing using depictive representations (as argued by
> Kosslyn).
> According to computationalists all internal processes can be explained
> by mindless mechanical procedures in which it is not the shape of the
> symbols which plays any importance, but the rules (algorithms) which
> are followed to achieve a result.

Close: The symbols can be interpreted as meaning something, but the
shape of the symbols is arbitrary, neither resembling nor causally
connected to what they mean. The rules are rules operating on those
symbol shapes; they are formal, or syntactic rules.

> They claim that the actual images
> we experience are epiphenomenal to these procedures. The strength of
> this argument is that computers using propositions seem to be able to
> do everything that depictions can do and also provide working models
> to corroborate their case.

Symbolic descriptions can come as close as you like to doing what
analog depictions do; and they can do things analog depictions cannot
do, or cannot do easily.

> Kosslyn himself readily admits in his
> 'Image and Brain' paper that "one can always form a propositional
> theory to mimic a depictive one" (pg.12.) It has been suggested,
> however, that this standpoint infers that what each of us is actually

implies, not infers

> aware of does not play any role at all and if this is the case, then
> one should ask why there is a mind present in each of us if mindless
> processes can do it all.

Good question; but what is the answer? (I don't expect you to have one,
but be aware that this question about awareness (it's the mind/body
problem again, of course), may not have an answer.

> The initial argument against images is that they cannot be directly
> measured or displayed and that there can be no little person in the
> brain (homunculus) to look at them, but Kosslyn was able to show
> through a computer program that analogue processing does not need a
> homunculus. In addition, by carrying out mental scanning experiments
> he was able to confirm his hypothesis that if mental imges are
> depictions (which unlike propositions retain the spatial properties
> of what they represent,) then more time is needed to move attention
> greater distances across imaged objects.

Would a kid-sib know what the last sentence had said?

> Furthermore, the use of
> depictions seems to make sense in the light of the neurophysiological
> discovery that area V1 in the brain is retinotopically mapped
> (preserving spatial relations) and that by using PET scans there have
> been comparisons for activation sites for both seen objects and
> mentally visualised objects (e.g. for size.)

And what was the result?

> If, as Pylyshyn argued, images are epiphenomenal, then damage in area
> V1 should not affect imagery. It has been found, however, that
> patients with hemispatial neglect also fail to report objects on that
> side in an imagined scene and that if part of the representation of
> the visual field is missing (e.g. in occipital lesion,) then the size
> of the mental visual field is reduced.
> Despite the above findings, the evidence for analogue processing in
> the brain is not conclusive. Supporters of symbol manipulation as a
> unitary amodal representational system have explained the mental
> scanning experiments in terms of lists of propositions and networks
> and there has been at least one clinical case of complete loss of
> visual imagery with preservation of normal visual perception.
> It can be seen then that there has not yet been a clear-cut winner of
> the imagery debate which looks set to continue until there is a
> compelling case which convinces everyone that one theory is correct
> and the other is incorrect. In the meantime I am happy to accept
> that mindless symbol manipulation is the best way to describe what is
> going on in my head (without ruling out the possibility that analogue
> processing might be used for visual and motor imagery and bearing in
> mind that although computers may be able to do what we do, it has not
> been definitively proved that the way they do it IS actually the way
> we do it.)

Good job!

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:42 GMT