Predictions Rule Perception: How Our Expectations Shape What We Perceive

People tend to trust that their perception of the world accurately reflects reality. However, work in psychology and neuroscience has suggested that our minds often deceive us, and our brains put a creative spin on the sensory information received from our senses. This means that two people looking at the same objective scene can have a different subjective experience of what it is they are perceiving. More surprisingly, our brains can even create the perception of something that is not there at all.

This phenomenon was demonstrated in our study. We conducted experiments where subjects were repeatedly shown two briefly presented stimuli simultaneously — a primary stimulus (e.g. a face) and a secondary stimulus (e.g. a square surrounding the face). Throughout the experiment, subjects maintained gaze fixation at the center of the screen, while the face-square stimulus pair flashed on the screen. After the briefly visible stimulus screen had disappeared from view, participants were either asked a question about the gender of the face (90% of the time) or, more rarely, a question about the subjective clarity of the square (10% of the time).

In general, people considered this task fairly simple, albeit tedious. What they did not realize, however, was that we had played a trick on them: occasionally, we omitted the less relevant stimulus altogether but still queried the subjects about it. In other words, in some cases, there was no square at all surrounding the face, but we nevertheless asked the participants to rate how clearly they perceived the square. You would think that participants immediately told us, “This is stupid, there was no square,” but alas, they did not. Results showed that most people reported having perceived the square even when it was not on the screen. We found that not only is this type of error possible, but it is also quite common, as over 90% of the subjects at least once reported some subjective experience of the absent stimulus. We also showed that this phenomenon can reliably be evoked across several analogous tasks.

This seems counterintuitive. Why would the subjects not notice the absence of the stimulus? Isn’t it important to accurately perceive what is out there and what is not? As a matter of fact, our results fit with many other findings from the field showing that our brains do not simply passively receive information from the environment; our brains generate and sometimes hallucinate this information. It appears that the brain actively creates our subjective reality based on each individual’s collection of knowledge, beliefs, and expectations. In this case, we can interpret our results to indicate that the perception of the square was created by a strong expectation to see it there.

This phenomenon has been explained by predictive coding theory, which posits that perception is always a mixture of prior knowledge and sensory data. Prediction signals are sent from higher cognitive areas of the cortex to lower hierarchical levels of vision, while sensory information moves upwards from lower primary visual areas. Generally, this is a part of optimal brain functioning to ensure seamless and economical processing of information gathered from the outside world. Errors only occur when predictions are given too much credence relative to the incoming information, likely due to strong habit or cumulative experience.

This is what happened in our experiment: we tricked the subjects’ brains to assume that this was a simple boring experiment where various faces appear together with squares. The situation was deemed predictable enough not to require any additional resources being allocated to meticulously study each repeated occurrence of the square. So in the rare instances when there was no square, the brain predicted its presence anyway and filled in the rest. By repeatedly presenting two stimuli together, expectation builds in such a manner that when one of those stimuli is removed, most people fail to notice and our brains fill in the gaps of the visual scene with elements which are most probable or expected to be present, thereby resulting in illusory perception. By the way, this phenomenon is not unique to our weird experiments. For example, researchers from Yale School of Medicine have demonstrated that subjects could hear tones that in reality were not there when they had been trained in a setting where visual images are always paired with auditory tones.

The same mechanism is thought to be at the root of various visual and auditory hallucinations. For instance, it has been suggested that people with schizophrenia experience hallucinations because the brain attributes too much confidence in predictions, which override the actual details present in the environment. In other words, people with a proneness to psychosis might be more inclined to interpret the world through lenses sculpted by their own expectations. Indeed, it has been demonstrated that schizophrenia patients with auditory hallucinations were more likely to hear tones that in reality were not there.

On the other hand, autistic traits have been linked with an overweighting of sensory information or weaker predictions. In that case, even the smallest differences in the environment can be interpreted as an entirely novel experience requiring detailed investigation, as opposed to more seamless perceptual processing where certain gaps are filled in by expectations. This explains why autistic people often feel overwhelmed and distressed in everyday situations, especially in a social context where there is a myriad of ever-changing details and cues. In our experiment, we also used a questionnaire to measure the autistic traits of participants. We found that people who scored higher on traits characteristic to autism (such as attention to detail) were less inclined to perceive the absent square in one of the tasks. This is in line with the hypothesis that autistic people in certain situations rely less on predictions than might be considered optimal. As a result, they are less susceptible to certain illusions, but may also get caught up in processing sensory details to an inordinate extent.

While studying irregularities in perception can hint at the neural mechanisms behind several clinical disorders, it also serves to enhance our still-developing understanding of normal perceptual processing. Our experiments illustrate that “hallucinating” some aspects of the visual scene is much more common than we might have previously thought. The predictive mechanisms of our brain are making sure that the infinite amount of information arriving through our senses is processed as smoothly and economically as possible, sometimes at the expense of objectivity. Although this means that we cannot always trust our perception, we can usually rely on our brains to get us through even the most tedious experiments with a touch of creativity.

Reference:

• Aru, J., Tulver, K., & Bachmann, T. (2018). It’s all in your head: Expectations create illusory perception in a dual-task setup. Consciousness and Cognition, 65, 197-208.