Interface theory of perception

[buket]

Do you have any opinion about 'interface theory of perception' by Donald Hoffman?
http://www.cogsci.uci.edu/~ddhoff/interface.pdf

 

[Solipsis]

An evolved organism shouldn't want to 'see the world as it is'. Indeed it's adapted to survive and its perception is finetuned to function optimally to serve that purpose, not to see things for what they truly are.

A designed organism like AI may give us in the future may be different. As long as a key motivator is to share our values to further mankind rather than harm us (for example to protect planet earth from us, as lofty a value as that might be to some dehumanized standards) it could probably be endowed with cleaner perception, but whenever you "want something" it probably makes perception necessarily selective.
In my actual experience, being too devoid of feeling makes it difficult to make even small choices since nothing is driving you either way. It's still possible to observe, but it really limits interaction or input/output. So to speak, you need some well-defined mission parameters but otherwise keep it clean. We humans have come a long way evolving but are absolutely chock full of vestigial artifacts from ages of evolution.

So yes I believe perception as we know it is heavily colored, seeing the world as it is is a heavily abstract concept because contemplating what is seen as we know it is so deeply intertwined with having feelings about it, desires, drives, etc. To optimalize an AI would on the one hand mean that such contemplation is not really present and any preference of function due to programmed values can be extremely important but also relatively marginal. On the other hand, you would use such an advanced AI to optimalize problem-solving, so that would mean you actually take advantage of this narrowing of scope and 'hyperfocus'.

I liked Hoffman's ted talk.

This thread is probably more a science & technology topic if you consider psychology enough of a science, rather than to say we philosophize on this matter. It doesn't really involve neuroscience but the mind. In my opinion anyway.

 

[buket]

An evolved organism shouldn't want to 'see the world as it is'. Indeed it's adapted to survive and its perception is finetuned to function optimally to serve that purpose, not to see things for what they truly are.

A designed organism like AI may give us in the future may be different. As long as a key motivator is to share our values to further mankind rather than harm us (for example to protect planet earth from us, as lofty a value as that might be to some dehumanized standards) it could probably be endowed with cleaner perception, but whenever you "want something" it probably makes perception necessarily selective.
In my actual experience, being too devoid of feeling makes it difficult to make even small choices since nothing is driving you either way. It's still possible to observe, but it really limits interaction or input/output. So to speak, you need some well-defined mission parameters but otherwise keep it clean. We humans have come a long way evolving but are absolutely chock full of vestigial artifacts from ages of evolution.

So yes I believe perception as we know it is heavily colored, seeing the world as it is is a heavily abstract concept because contemplating what is seen as we know it is so deeply intertwined with having feelings about it, desires, drives, etc. To optimalize an AI would on the one hand mean that such contemplation is not really present and any preference of function due to programmed values can be extremely important but also relatively marginal. On the other hand, you would use such an advanced AI to optimalize problem-solving, so that would mean you actually take advantage of this narrowing of scope and 'hyperfocus'.

I liked Hoffman's ted talk.

This thread is probably more a science & technology topic if you consider psychology enough of a science, rather than to say we philosophize on this matter. It doesn't really involve neuroscience but the mind. In my opinion anyway.

I think there should be some resemblance between reality and our perception of it.. Do you think Hoffman has evidence to prove his theory to be true?

 

[Solipsis]

Of course there is resemblance. Remember that the theory is that our brains interpret, guess and fill in, but what they fill in is the limited sensory information we get in favor of making effective choices. Actually I think hallucinations are also relevant here, the kind we have sober and that happen to virtually everyone - I am talking about mistaking a gardenhose for a snake for example but in such a way that we really thought we saw a snake for a minute. Yes this can occasionally involve hallucinations and it is normal and not unhealthy.
The reason or explanation being that there is an evolutionary advantage to having a very fast reaction to potential dangers, and it's better to be wrong a few times too many than to miss a real threat, especially to survival it is.

Obviously we don't just see random things that have nothing to do with physical reality... our senses are still part of the picture and his theory doesn't dispute that. But the flexible nature of our perception can hardly be disputed.

The part of his theory about perception being nothing like fundamental physical reality has to do with objectivity vs subjectivity and certain irreconcilable differences between how we observe classical physics and what quantum mechanics is like. This also can hardly be disputed: yes we only see a very selective frame at one given time, and everything about that is optimized for our functioning and not for true analysis or objective observation. The evidence to support it lies in the correlations between evolutionary game theory and ways our mind works. The hallucinations I mentioned is one example of how we evolved to survive and how plenty often this goes at the cost of 'unadulterated perception'.
Like I argued before, a much more 'objective' perception may be pretty dysfunctional or even unworkable, unless you would tweak it in really smart ways, and that is not how evolution do.

Why should there be more than some limited resemblance between reality and our perception of it, beyond what the manifestations of physics that are relevant to us cover? We have found that there is a huge electromagnetic spectrum, but the visible light spectrum is but a tiny slice of that spectrum that has just what we need. It is real, but to think it's a fairly complete view of reality is a very skewed view.
Another thing to remember is that we are so trapped in our human perception that we shouldn't expect to have any initial reaction besides outrage or ridicule of how or why much of reality would really be that obscured. It suits us just fine, and it's often overwhelmingly much already. But that is not necessarily what you then should believe. Like the rest of the electromagnetic spectrum, invisible to us but certainly there and occasionally manifesting, proving to be real like X-rays, it's good to contemplate that we are humble.

imho

P.S. considering what mankind has achieved though, it helps to discuss this in relative terms. Yes using tricks we can overcome our limits and practice science, philosophy and other things, but I guess this theory looks at those limitations pretty harshly.