Re: Imagery Debate

From: Harnad, Stevan (
Date: Sun Feb 04 1996 - 14:46:20 GMT

> From: "Parker, Chris" <>
> Date: Sun, 4 Feb 1996 14:31:55 +0000
> The Imagery Debate is all about whether the images that we all know and
> experience through introspection are useful entities which we can
> investigate and ground in theories, or whether they are really
> something of a private illusion that has no formal role in anything or
> at best a minor role. We believe that there have to be some form of
> representations in our head, but are any of them in the form of an
> image?

What are "representations"? We all know we have the experience of a
seeing-like event under certain conditions when we are not actually
looking at the object. That's a Cartesian fact about our mental lives.
But that's just appearances: What is actually going on in our brains?
How do we "use" these images, if we use them at all: In fact, who/what
are "we"? We can't just be little men in our own heads (homunculi)
looking at the images the way we look at objects in the world -- for
otherwise the question switches to what is going on in the heads of the
little men in our heads...

> The homunculus fallacy is all about the conferring of human attributes
> to something that is not human or is only part of a human and using
> that something as an actor in a process. Not only is this incoherent,
> but all that is added to any explanation is an extra unhelpful step. If
> we are driven by an homunculus, who drives the homunculus.?
> An extreme case would be to describe a process as our mind seeing and
> scanning pictures on our visual cortex and telling our legs to start
> running, a less obvious case would be to say that the brain interprets
> and understands messages sent to it by our sensory organs. Metaphors
> should be acknowledged for what they are and never extrapolated to
> become real entities.

Good. There is also a causal load that such "explanations" are incapable
of carrying...

> Pylyshyn criticises the use mental imagery for memory representations,
> suggesting that it is adequate to propose abstract mental structures or
> symbolic descriptions which are not accessible to our consciousness. He
> suggests that propositions are likely to be more robust representations
> than appearances of images, and that representations of images are more
> like descriptions than pictures. This suggests that our introspection
> of an image is something of a construct. Stored images would need to be
> scanned by something, a set of propositions would not, they could be
> part of an information processing model.

You should say something about what computation is here (symbol
manipulation) so kid bro knows what would actually be going on in his
head and in his computer if Pylyshyn was right. But then explain why the
manipulation of analogue images could not be going on in there just as
readily as the manipulation of symbols, if it is sufficient to get the
job done (and could be done by a machine without the help of any

> Kosslyn defends imagery by acknowledging that the primitive picture
> metaphor should be rejected because of the need for a second, third etc
> processing system to interpret the information from the previous
> process, but pleading the case for quasi-pictorial images as a "special
> kind of functional representation in human memory".

What does "quasi-pictorial" mean? In general, the arguments for and
against imagery and computation need more substance and examples,
otherwise kid-bro walks out bemused...

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:58 GMT