Perception – direct perception and philosophy

by Neil Rickert

In an earlier post, I wrote “a proper understanding of direct perception actually tends to undermine both traditional epistemology and traditional philosophy of mind.”  Today, I want to expand on that.

Both traditional epistemology and traditional philosophy of mind assume that humans are rational agents.  So let’s take a look at what is a rational agent.

Rational agents

An account of the basic ideas of rational agency can be found in the Wikipedia entry, or in the Stanford encyclopedia.

The basic idea is that perception presents the state of the world to a rational agent.  The agent then uses logic in order to decide which actions will best meet the agent’s desires.  If the logic is too complex (or computationally intractable), the agent might instead use an heuristic which approximates what should be the result of that logical deduction.

It is, in short, an input/output model.  The agent receives input, and computes appropriate actions (outputs).  This kind of input/output model fits the way that we use computers, and is the basis for a lot of research on Artificial Intelligence.  This input/output model is also behind the commonly expressed view that the brain is a computer.

It is part of the typical rational agency view, that perception is passive.  The general view is that sensory cells pick up signals from the world (or the environment), and determine everything important from those passively received signals.  In the view of many, the inputs themselves are meaningless as received, and any meaning is added by the computation that the brain does on the passively received inputs.

Active perception

It is part of the view of the proponents of direct perception, that perception is active, not passive.  We are part of the world, and as such we continuously interact with the world.  We are not outsiders looking on.  Gibson points to the motion of the eyes in saccades, as evidence of our active engagement in the perceptual process.  This emphasis on our interaction with the world is sometimes called interactionism or embodied cognition.

Gibson describes his account of perception as an ecological theory.  The name reflects the interactive nature of perception.  He says that we perceive affordances, where an affordance is something that affords us or offers us an opportunity.  A mail box affords us the possibility of being able to mail a letter.  Food affords us the opportunity of obtaining nourishment.  I prefer to express that by saying that we perceive opportunities – the word “opportunity” is less confusing than the word “affordance.”

For me, the idea that we perceive opportunities comes from the assumption that we are the products of biological evolution.  It seems implausible that evolutionary processes would result in something like the rational agent of a Platonic ideal.  Rather, evolutionary processes should be successful to the extent that they produce organisms or agents that are able to meet their biological needs.  So this evolutionary approach suggests that perceptual systems should have evolved to identify opportunities to succeed.  For this, the term “opportunistic agent” seems to be a better fit than “rational agent.”

If I do a google search for “opportunistic agent”, I find references to such things as weeds or disease causing microbes and parasites.  The idea of biological systems being opportunistic is well understood.  As biological organisms ourselves, it makes sense that we too should be opportunistic.

The term “opportunism” is often seen as a negative, perhaps suggesting an amoral greed.  However, there is no reason that it cannot also be seen in a more positive light, a seeking of moral opportunities or opportunities to help our fellow citizens as members of a social group.

If perception seeks opportunities, as suggested, then what we perceive is already meaningful to us, by virtue of it satisfying our needs.  So we at least have a beginning of a way of understanding where meaning originates.

Perceptual learning

If what we perceive are opportunities, then part of learning is in learning what are the environmental indicators of opportunities.  This is perceptual learning.  On Gibson’s account, we build transducers to recognize particular types of opportunity.  So a great deal of our learning is in the form of perceptual learning.

I remember, in my early teens, going with a group (mainly adults) on a kangaroo hunt near Busselton, Western Australia.  One of the participants pointed, and said “there’s one.”  I looked in the direction that he was pointing, but I did not see a thing.  It turned out that the kangaroo was partially hidden in the bushes.  Shortly thereafter, somebody aimed, and shot himself a kangaroo.  From that experience, I learned that how to see kangaroos (or other game) is something that we have to learn.  How to properly see what is under a microscope is another example of having to learn.  These are examples of perceptual learning, of tuning our perceptual systems to be able to pick up the kind of information that we are seeking to present us with opportunities.

One of the things that a child must learn, is how to cross safely cross the street.  Presumably, the “rational agent” view of this would be that the child must absorb many rules of safe crossing, and then use logic to apply those rules to the current situation.  The “opportunist agent” view is that the child learns to recognize opportunities to safely cross the street, and to choose such an opportunity to cross.

Epistemology

Typically, “epistemology” is said to be a theory of knowledge.  And it is often said that knowledge consists of justified true beliefs.  If we receive neutral data from the world, and what we know of the world comes from applying logic to that neutral data, it makes sense to say that knowledge is in the form of beliefs.

From the perspective of direct perception, that is not how we come to know the world.  Rather, much of our learning is in the form of perceptual learning, or of tuning up the perceptual system to recognize opportunities that are available to us.  Thus we should see knowledge as residing in that perceptual ability, as resulting from that perceptual learning.

This suggests that knowledge and belief complement one another.  Knowledge provides us with the abilities to recognize opportunities, and then beliefs are the information on specific opportunities that have been recognized.

One of the problems of epistemology, is that of underdetermination, sometimes called “the poverty of stimulus problem.”  If our only access to the world is via a finite set of observed facts, then there are infinitely many possible worlds that could fit those facts.  So how could we possibly be able to tell what the world is like?  This is a limitation of the idea that perception is passive, so that we only have access to data received from the world.  If perception is active, as direct perception proposes, then we have both data from the world (or beliefs), and the learned methods by which we acquire that data (our knowledge).  Having available the method of acquiring meaningful data, in addition to the data itself, tells us far more about the world than would just the data.  This takes us in the direction of solving the underdetermination problem.

Summary

I have described some of the ways that direct perception gives us a picture of our relation to the world that is very different from what is presented by traditional philosophy.

Advertisements

2 Comments to “Perception – direct perception and philosophy”

  1. “If our only access to the world is via a finite set of observed facts, then there are infinitely many possible worlds that could fit those facts. So how could we possibly be able to tell what the world is like? This is a limitation of the idea that perception is passive, so that we only have access to data received from the world. If perception is active, as direct perception proposes, then we have both data from the world (or beliefs), and the learned methods by which we acquire that data (our knowledge). Having available the method of acquiring meaningful data, in addition to the data itself, tells us far more about the world than would just the data. This takes us in the direction of solving the underdetermination problem.”

    I tend to agree with Churchland’s or even Rock’s ideas over those of Gibson. I believe that perception is active rather than passive, and this reinforces my view that perception is indirect rather than direct, or at the very least, it is a combination of the two. That is, I agree with you that the body and brain are finely tuned to see certain things in the environment based on evolution over time increasing a beneficial view of reality (let’s say a view that is more closely correlated with the world as it is “out there”), thus providing us with perceived “opportunities” that may seem direct in some sense. However, to negate cognitive processes in the construction of those percepts that constitute our reality seems fallacious to me. We see what we want to see and Gestaltism has demonstrated that we perceive with a top-down approach as well as bottom up, which implies that we modify or interpret sensed data differently, that is, we often perceive the world in a holistic way which depends on the patterns our brain is programmed/trained to look for. Even with the same brain and tuned transducers, the interpretation changes with emotional states, and thus our view of reality is dynamic in ways that don’t exclusively better match it — thus implying that some form of representationalism is going on with a certain amount of flexibility in combination with environmentally tuned transducers. If we are really seeing things as they are, then how do we account for things like optical illusions that are largely based on this “gap filling” or Gestalt mechanism which ends up creating things or effects that we believe are not actually there (illusion of objects moving, having incorrect sizes, etc.). It seems to me that a combination of direct and indirect, which we could just call indirect (with a certain amount of correlation with things as they are “out there”, enough to promote decent survival rates) seems a more plausible view of perception. That’s just my opinion. I do think that there are quite a few compatibles between the two theories however (indirect and direct).

    I’m not sure what you specifically mean by neutral data. To me “neutral data” implies just “data”, as I see all data as being neutral in that it doesn’t carry objective meaning along with it — for that is up to the interpreter, just as is the case with any scientist performing an experiment. The data is there, but any conclusions drawn or meaning behind the data is based on how it is interpreted. Perhaps you mean something different by “neutral”, that I’m not considering. It’s a fine line drawn (and arbitrary in at least some sense) separating direct from passive perception. Acquiring meaningful data could be seen as producing the same result as acquiring “neutral data” and applying meaning to it via the context our brain superimposes on that neutral data — where the context or meaning is ingrained via evolved “opportunistic” mechanisms that happen to increase survival. It seems like a gray area with a lot of overlap though.

    Even if direct perception was the case, I disagree that it would solve, or takes us in the direction of solving, the underdetermination issues that Quine and others have discussed (when applied to epistemology in general). There are still infinitely many worlds that could fit any direct perceptions we may be experiencing because we can never hope to have perfect “transducers” (thus leaving gaps between what we perceive and what the world could really be like if independent of our minds) among other reasons. I see no ultimate difference between seeing the world incompletely via imperfect direct perception or indirect representationalism. Either could have evolved as long as they promote survival enough to overcome any shortcomings.

    Although I don’t completely agree, I do like this post.

    Like

    • I believe that perception is active rather than passive, …

      Good.

      …, and this reinforces my view that perception is indirect rather than direct

      I was not expecting to persuade everyone.

      I have not yet gone into details of what distinguishes indirect perception from direct perception. That will come in a future post.

      However, to negate cognitive processes in the construction of those percepts that constitute our reality seems fallacious to me.

      Keep in mind that I am equating perception to the obtaining of information about the environment. There can be cognitive processes involved in using that information.

      Like

%d bloggers like this: