donald_d_hoffman's picture
Cognitive Scientist, UC, Irvine; Author, The Case Against Reality
Truer Perceptions Are Fitter Perceptions

Those of our predecessors who perceived the world more accurately enjoyed a competitive advantage over their less-fortunate peers. They were thus more likely to raise children and to become our ancestors. We are the offspring of those who perceived more truly, and we can be confident that our perceptions are, in the normal case, reasonably accurate. There are of course endogenous limits. We can, for instance, see light only in a narrow window of wavelengths between roughly 400 and 700 nanometers, and hear sound only in a narrow window of frequencies between 20 and 20,000 Hertz. Moreover we are prone, on occasion, to have perceptual illusions. But with these provisos noted, it is fair to conclude on evolutionary grounds that our perceptions are, in general, reliable guides to reality.

This is the consensus of researchers studying perception via brain imaging, computational modeling and psychophysical experiments. It is mentioned in passing in many professional publications, and stated as fact in standard textbooks.

But it gets evolution wrong. Fitness and truth are distinct concepts in evolutionary theory. To specify a fitness function one must specify not just the state of the world but also, inter alia, a particular organism, a particular state of that organism, and a particular action. Dark chocolates can kill cats, but are a fitting gift from a suitor on Valentine's Day. 

Monte Carlo simulations using evolutionary game theory, with a wide range of fitness functions and a wide range of randomly created environments, find that truer perceptions are routinely driven to extinction by perceptions that are tuned to the relevant fitness functions. The extension of these simulations to evolutionary graphs is in progress, and the same result is expected. Simulations with genetic algorithms find that truth never gets on the stage to have a chance to go extinct.

Perceptions tuned to fitness are typically far less complex than those tuned to truth. They require less time and resources to compute, and are thus advantageous in environments where swift action is critical. But even apart from considerations of time and complexity, true perceptions go extinct simply because natural selection selects for fitness not truth. 

We must take our perceptions seriously. They have been shaped by natural selection to guide adaptive behaviors and to keep us alive long enough to reproduce. We should avoid cliffs and snakes. But we must not take our perceptions literally. They are not the truth; they are simply a species-specific guide to behavior. 

Observation is the empirical foundation of science. The predicates of this foundation, including space, time, physical objects and causality, are a species-specific adaptation, not an insight. Thus this view of perception has implications for fields beyond perceptual science, including physics, neuroscience and the philosophy of science. The old assumption that fitter perceptions are truer perceptions is deeply woven into our conception of science. The funeral of this assumption will not be snubbed with a back-page obituary, but heralded with regime change.