Interface Theory of Perception

saor

Honorary Master
Joined
Feb 3, 2012
Messages
20,120
A quick summary of the idea, original paper here:
http://cogsci.uci.edu/~ddhoff/Hoffman-Stevens-Handbook.pdf

Our perception of the world is a product of evolution in the same way that our legs and eyes are. It's been assumed that natural selection has shaped our perception to favor accurate representations of the world. The tiger that most accurately sees the world has the most advantage / fitness in that world.

What the objective world is like:
download (1).jpg

What the tiger sees (or some close approximation thereof):
download (1).jpg

Some quotes:
● “Evolutionarily speaking, visual perception is useful only if it is reasonably accurate ...
● “Visual perception ... involves the evolution of an organism’s visual system to match the structure of the world and the coding scheme provided by nature.”
● “In general, (perceptual) estimates that are nearer the truth have greater utility than those that are wide of the mark.”
Now, enter the world of simulations of evolutionary game theory. What wins in these games isn't perception that's tuned toward accurate estimations of reality, but rather perception that provides maximum fitness; perception doesn't care about accurate representations of reality - it cares more about fitness. What might this mean? Well...consider your desktop:

desktop.png

When I interact with my computer I do so via the desktop interface. I manipulate icons. These icons are not accurate representations of the binary and circuitry beneath the surface; they are simple abstractions which allow me to interface with a complex system. Indeed - I don't need to manipulate voltages and code to write a document - I just need the interface with this interface and it's symbols that allows maximum fitness in this world of computer interactions.

And so in a similar way, our perception of reality is a 'desktop interface' that uses (fitness-tuned) abstractions to make sense of a more complex reality. In this schema the perception of the tiger would be more like this:

What the objective world is like:
bw.jpg

What the tiger sees:
download (1).jpg

I found the framing of this in terms of a desktop interface to be quite interesting: We have a context for understanding how simplifications and abstractions allow us to better interact with computers whilst also being aware that those abstractions bear no correlation to the underlying reality of the computer. Might our perceptions of reality be the same - a species specific desktop interface that bears no resemblance to what reality is really like?
 

Xarog

Honorary Master
Joined
Feb 13, 2006
Messages
19,041
I found the framing of this in terms of a desktop interface to be quite interesting: We have a context for understanding how simplifications and abstractions allow us to better interact with computers whilst also being aware that those abstractions bear no correlation to the underlying reality of the computer. Might our perceptions of reality be the same - a species specific desktop interface that bears no resemblance to what reality is really like?
Consider carefully: Every attempt to objectively measure time is in fact a measurement of two spaces. If our perception of time is wrong (how likely is that given that it is a purely internal sense?), what hope do we have of correcting it?

Also: The thing that makes you afraid of the tiger when it appears before you isn't actually to be found in/on the tiger. We could not actually interact with the world without our prejudices. Which is to say, we live in a reality augmented by our interface and it's better that way.
 

Bobbin

Executive Member
Joined
Oct 22, 2009
Messages
6,312
I think I stumbled on a similar idea somewhere in PD before. Talking about how we are adapted to detect particles through our senses when particles don't actually exist but are rather just a certain combination of waves (Constructive or destructive interference). But how our reality is so shaped by our perception of particles - that our mind actually "creates" the particles through perception. i.e. the eye detecting and interpreting a single photon.

It is the only way my silly brain made sense of one aspect of quantum theory, by rejected the idea of particle/wave duality alltogether and just believing that everything is made of only waves :p

You kind of get into a weird state when you look down at your desk and start pressing on it, wondering what color it really is (If any) and what forces are pushing back on your hand when you aren't truly touching the desk. :D
 
Last edited:

saor

Honorary Master
Joined
Feb 3, 2012
Messages
20,120
You kind of get into a weird state when you look down at your desk and start pressing on it, wondering what color it really is (If any) and what forces are pushing back on your hand when you aren't truly touching the desk. :D
I did the same thing.
Mind kept trying to think about what reality really is if the tiger and chairs are merely icons on my interface.

A slight objection / question I had about the paper was this:

The entire ITP theory is based on Monte Carlo simulations of evolutionary games in which fitness beats truth, aka: Those that come out winners are those who's perceptual systems favour fitness (a good interface) rather than those which favour truth (accurate representations of reality). Isn't this making the assumption that evolution is an optimal engine of design? Surely you could run those simulations for say, different eye configurations and get results which prove that 3 eyes provides for optimal fitness. But of course - evolution doesn't always maximise fitness absolutely. How then do you construct a theory like ITP based on the assumption that evolution is always absolutely optimal?

Also: The thing that makes you afraid of the tiger when it appears before you isn't actually to be found in/on the tiger. We could not actually interact with the world without our prejudices. Which is to say, we live in a reality augmented by our interface and it's better that way.
To be clear: In ITP the tiger is the icon; human reality is the interface. We have no access to 'ultimate reality'. When you say 'augment', I take you to mean we perceive reality but our response to it is shaped by our cognitive biases and prejudices.
 
Last edited:

Xarog

Honorary Master
Joined
Feb 13, 2006
Messages
19,041
I did the same thing.
Mind kept trying to think about what reality really is if the tiger and chairs are merely icons on my interface.

A slight objection / question I had about the paper was this:

The entire ITP theory is based on Monte Carlo simulations of evolutionary games in which fitness beats truth, aka: Those that come out winners are those who's perceptual systems favour fitness (a good interface) rather than those which favour truth (accurate representations of reality). Isn't this making the assumption that evolution is an optimal engine of design? Surely you could run those simulations for say, different eye configurations and get results which prove that 3 eyes provides for optimal fitness. But of course - evolution doesn't always maximise fitness absolutely. How then do you construct a theory like ITP based on the assumption that evolution is always absolutely optimal?

To be clear: In ITP the tiger is the icon; human reality is the interface. We have no access to 'ultimate reality'.
Here is the real question. How much energy does it cost an organism to effect an accurate representation of reality versus one that merely effects effectiveness?

I think the former is far more expensive in terms of time and energy. Your brain has a limited capacity and it is by far the most energetically expensive tissue in the body, so investing in brain matter comes at a premium. So I would say that it's not so much a bet that evolution is an optimal design system so much as it is that evolution is a design system which favours removing the least-optimal designs from the system. It seems to me that the less energy you need to think, the more energy you have left over to act.

And isn't it convenient that the effective version of the system is optimised (in theory) to simply tell you how you need to act in the world rather than wasting energy on telling you what the world really is like?
 

saor

Honorary Master
Joined
Feb 3, 2012
Messages
20,120
Here is the real question. How much energy does it cost an organism to effect an accurate representation of reality versus one that merely effects effectiveness?
Isn't there a chance the energy cost is dependent on the nature of reality?

A.) If macroscale objective reality = a thing that's shaped like a tree, what's the energy cost of truth vs. fitness,
B.) If objective reality = a cloud of strange fluctuations, what's the energy cost of truth vs. fitness

In the case of A maybe it is just the case that truth incurs a lower energy cost.
In the case of B maybe it's the case that fitness incurs a lower energy cost. Although I suspect with B the energy cost might not be a consideration - the need for fitness is imposed rather than offered. Unless there's a way of being in the world of B that allows for truth rather than fitness. Of course - it's impossible to conceive of such a way of being.

And isn't it convenient that the effective version of the system is optimised (in theory) to simply tell you how you need to act in the world rather than wasting energy on telling you what the world really is like?
Agreed, and it's a nice consideration to the issue I was having with evolution being always optimal. One can almost imagine truth slowly being chipped away by evolution in favour of fitness and how to act; in the beginning there was light, but that light slowly fades so that we might better traverse the world.
 
Last edited:

Xarog

Honorary Master
Joined
Feb 13, 2006
Messages
19,041
Isn't there a chance the energy cost is dependent on the nature of reality?

A.) If macroscale objective reality = a thing that's shaped like a tree, what's the energy cost of an truth vs. fitness,
B.) If objective reality = a cloud of strange fluctuations, what's the energy cost of truth vs. fitness

In the case of A maybe it is just the case that truth incurs a lower energy cost.
In the case of B maybe it's the case that fitness incurs a lower energy cost. Although I suspect with B the energy cost might not be a consideration - the need for fitness is imposed rather than offered. Unless there's a way of being in the world of B that allows for truth rather than fitness. Of course - it's impossible to conceive of such a way of being.
I don't really think it matters what the nature of reality is. My argument is basically this: If we're assuming any sort of external existence aside from our consciousness, then it presumably operates according to a set of rules. To build a real model of this external reality requires that your representation of reality contains all the information that is really available. Presumably there is an energy cost associated with maintaining a respresentation of any sort of information, regardless of whether or not evolution is the principle by which life perpetuates itself or not.

That is to say, it seems clear that not all information is necessarily relevant to what you are doing at any given time. It seems far more efficient to reduce the information you're dealing with to only that which you need in order to act "successfully" in the world.

Agreed, and it's a nice consideration to the issue I was having with evolution being always optimal. One can almost imagine truth slowly being chipped away by evolution in favour of fitness and how to act; in the beginning there was light, but that light slowly fades so that we might better traverse the world.
Yes. And instead of the memory, you inherit the instinct, because the instinct is the refinement of the memory.
 

DMNknight

Expert Member
Joined
Oct 17, 2003
Messages
3,257
We've had a similar discussion before, just on the nature of how the brain is captive to it's perceptions but is also completely dependent on them for survival, which has worked to date, so it can't be all that wrong.

if you break down the components of each sense we have, you get a real appreciation of what your sense do.
1) Eyesight - feel free to google the visible light waves that can be seen by the eye. biological photoreceptors that turn light into electrical signal for interpretation by the brain
2) Hearing - environmental vibrations which are essentially molecular/atomic kinetic excitation translated into electric signals
3) Smell and taste (very closely associated) - translating molecules present into electric signals
4) Touch - close to hearing because you can feel sound, but also allows you to "vibrate" against external matter in order to determine it's nature.

When you take all that you have into account, it is logical that your survival is predicated on these senses and that nothing that can kill us outside of these senses is currently in our environment. (within context of propagation of the species)
We may well encounter an alien species that is light based outside our visible spectrum *shrug*

It is sobering and humbling to realise how much we can do with so little.
 
Top