On Perception

From Edge Questions, Donald Hoffman writes about how he changed his mind about perception:

                  Veridical Perception              

I have changed my mind about the nature of perception. I thought                 that a goal of perception is to estimate properties of an objective                 physical world, and that perception is useful precisely to the                 extent that its estimates are veridical. After all, incorrect                 perceptions beget incorrect actions, and incorrect actions beget                 fewer offspring than correct actions. Hence, on evolutionary                 grounds, veridical perceptions should proliferate.

             

Although                 the image at the eye, for instance, contains insufficient information                 by itself to recover the true state of the world, natural selection                 has built into the visual system the correct prior assumptions                 about the world, and about how it projects onto our retinas,                 so that our visual estimates are, in general, veridical. And                 we can verify that this is the case, by deducing those prior                 assumptions from psychological experiments, and comparing them                 with the world. Vision scientists are now succeeding in this                 enterprise. But we need not wait for their final report to conclude                 with confidence that perception is veridical. All we need is                 the obvious rhetorical question: Of what possible use is non-veridical                 perception?

             

I now think that perception is useful because it is not veridical.                 The argument that evolution favors veridical perceptions is wrong,                 both theoretically and empirically. It is wrong in theory, because                 natural selection hinges on reproductive fitness, not on truth,                 and the two are not the same: Reproductive fitness in a particular                 niche might, for instance, be enhanced by reducing expenditures                 of time and energy in perception; true perceptions, in consequence,                 might be less fit than niche-specific shortcuts. It is wrong                 empirically: mimicry, camouflage, mating errors and supernormal                 stimuli are ubiquitous in nature, and all are predicated on non-veridical                 perceptions. The cockroach, we suspect, sees little of the truth,                 but is quite fit, though easily fooled, with its niche-specific                 perceptual hacks. Moreover, computational simulations based on                 evolutionary game theory, in which virtual animals that perceive                 the truth compete with others that sacrifice truth for speed                 and energy-efficiency, find that true perception generally goes                 extinct.

             

It                 used to be hard to imagine how perceptions could possibly be                 useful if they were not true. Now, thanks to technology, we                 have a metaphor that makes it clear — the windows interface of                 the personal computer. This interface sports colorful geometric                 icons on a two-dimensional screen. The colors, shapes and positions                 of the icons on the screen are not true depictions of what they                 represent inside the computer. And that is why the interface                 is useful. It hides the complexity of the diodes, resistors,                 voltages and magnetic fields inside the computer. It allows us                 to effectively interact with the truth because it hides the truth.

             

It                 has not been easy for me to change my mind about the nature                 of perception. The culprit, I think, is natural selection.                 I have been shaped by it to take my perceptions seriously.                 After all, those of our predecessors who did not, for instance,                 take their tiger or viper or cliff perceptions seriously had                 less chance of becoming our ancestors. It is apparently a small                 step, though not a logical one, from taking perception seriously to taking it literally.
               
                Unfortunately our ancestors faced no                 selective pressures that would prevent them from conflating                 the serious with the literal: One who takes the cliff both                 seriously and literally avoids harms just as much as one who                 takes the cliff seriously but not literally. Hence our collective                 history of believing in flat earth, geocentric cosmology, and                 veridical perception. I should very much like to join Samuel                 Johnson in rejecting the claim that perception is not veridical,                 by kicking a stone and exclaiming "I refute it thus." But even                 as my foot ached from the ill-advised kick, I would still harbor                 the skeptical thought, "Yes, you should have taken that                 rock more seriously, but should you take it literally?"