Neuroscience Shouldn’t Divorce Perception and Reality

by Joseph Shieber

There is a spate of popularizations of neuroscience promoting the idea that “reality isn’t something you perceive, it’s something you create in your mind”, that “everything we perceive is a hallucination created by the brain”, or — as one Scientific American article put it — “It is a fact of neuroscience that everything we experience is a figment of our imagination.”

These popularizations are unfortunate, because they run together at least two claims that we should distinguish from each other. The first claim is that perception isn’t merely a process in which the brain passively receives impressions from the world. Rather, it’s a process in which the brain is an active participant, testing predictions about the world against the inputs that it receives. The second claim is that, if the brain is an active participant in perception, then perception “creates reality in your mind” or is a “hallucination”.

Now, when neuroscientists make the first sort of claim, they’re speaking on the basis of their expertise as scientists. Indeed, the past thirty years have seen the rise of views in psychology and neuroscience that understand the brain as, in the words of  the noted philosopher and cognitive scientist Andy Clark, a “prediction engine”.

Here’s how Lisa Feldman Barrett puts this idea in her book, How Emotions are Made: The Secret Life of the Brain :

The discovery of simulation in the late 1990s ushered in a new era in psychology and neuroscience. What we see, hear, touch, taste, and smell are largely simulations of the world, not reactions to it. … Simulations are your brain’s guesses of what’s happening in the world. In every waking moment, you’re faced with ambiguous, noisy information from your eyes, ears, nose, and other sensory organs. Your brain uses your past experiences to construct a hypothesis—the simulation—and compares it to the cacophony arriving from your senses. In this manner, simulation lets your brain impose meaning on the noise, selecting what’s relevant and ignoring the rest. (Feldman Barrett, Chap 2)

Of course, it’s important not to downplay the significance of this development in our understanding of perception. Traditionally, philosophers drew a stark distinction between perception and inference. On this traditional view, perception is the primary route by which we acquire new information, whereas inference can only transmit information that we already acquired.

Furthermore, philosophical tradition has emphasized deductive inference, inference that relies on logically valid rules rather than inductive inference, inference that relies on mere probabilistic evidence.

Interpreting perception as prediction upsets both of these traditional tenets. If the perceiving brain is a prediction engine, the motor of that engine is probabilistic — and particularly Bayesian — inference. So not only does this new idea from cognitive psychology and neuroscience erase the distinction between perception and inference, but it also upsets the hierarchy of deductive over inductive inference so central to traditional philosophy.

In other words, the first claim — that perception involves the brain’s actively making and testing predictions about the world — is both squarely within the purview of the brain sciences and an original and interesting claim.

What about the second claim — that, if perception is constituted by the brain’s actively making and testing predictions about the world, then perception “creates reality in your mind” or is a “hallucination”? Unfortunately, this claim is neither original nor interesting — nor is it properly within the expertise of those brain scientists whose popularizations fuel its spread.

It’s not properly within the expertise of the brain scientists, because it involves concepts beyond the purview of those scientists — most importantly, the notion of “reality”.

It’s not original, because it is nothing more than a rehash of philosophical positions whose development goes back centuries, at least. If the view is that there is no reality at all beyond what exists in your mind, then it is a version of Berkeley’s idealism. If the view is that there may well be such a reality, but that our only grasp of it is through the structure imposed by our minds, then it is a version of Kant’s marriage of “empirical realism” and “transcendental idealism”.

The biggest problem with the second claim, however, is simply that it’s not terribly plausible. According to the second claim, the fact that the perception involves the brain’s hypotheses or predictions about the world means that perception doesn’t give us access to reality.

It strikes me that those popularizers who try to capture attention with overblown claims about the disconnect between perceptual experience and the “real” nature of the world are employing an unreasonable conception of what “access to reality” would have to involve.

If I accurately capture some features of the world around me, but in doing so fail to capture certain other features, have I achieved “access to reality”? It strikes me that I have — at least partially — particularly if those features of the world around me were the features I intended to capture or were the features that were salient to achieving the goals that I was pursuing.

Here’s a bit more evidence for the implausibility of the second claim. Can we perhaps think of other areas that involve hypotheses or predictions about the world? What about — just spitballing here — science itself?

Science itself is a collection of methods involving establishing hypotheses and using those hypotheses to make predictions about the world. It involves testing those predictions against the world and using the results of those tests either to invest greater confidence in the hypotheses or to revise the hypotheses, if the world didn’t turn out as the predictions suggested. Indeed, the development of the “predictive brain” paradigm used scientific hypothesis-testing as a model for understanding the brain as a prediction engine. As the neuroscientists Karl Friston and Klaus Stephan write, “Many people now regard the brain as an inference machine that conforms to the same principles that govern the interrogation of scientific data”. (For a dissent, see here.)

If some brain scientists are suggesting that a method’s involving hypotheses or predictions about the world means that the method doesn’t give us access to reality, then that thesis would apply to those scientists’ own method. Which means that their own thesis is self-defeating: we would have no reason to think that their own claims about the reality of the mind and its relation to the world are likely to be true.