My phone's alarm clock woke me up this morning. I looked in the direction of the music, found my phone on the bedside table, and reached to grab it. I was instantly surprised by its light weight, and realized that instead of my brick-like iPhone 6 (which I keep protected in a heavy waterproof, childproof case), I was holding a pocket notebook of roughly equal size.

Visual perception is thought to encompass not only the sensory input arriving to our retinas, but also our expectations about the nature of our surroundings. This means that perceptual errors can occur when sensory entries are noisy or weak, and also when we have erroneous expectations about the objects we expect to encounter in our visual environment.   

My misperception this morning is a case in point: I was sleepy, the light in the bedroom was dim (so the sensory input was weak), and I was strongly anticipating to find a phone on my bedside table. Had the light been more intense, or my expectation less so, I might not have made the mistake.

In a study published last September, neuroscientists Auréliane Pajani, Peter Kok, Sid Kouider, and Floris P. de Lange set out to investigate whether spontaneous neural activity patterns (which may indicate prior expectations) in the primary visual cortex of the brain might predispose an observe to experience hallucinations (false perceptions in the absence of a sensory signal). To answer this question, the team used fMRI (functional magnetic resonance imaging) to scan the brains of study participants while they indicated whether gratings having specific orientations (slants) were present on a computer screen.

The experimental setup was such that it predisposed the subjects to experience hallucinations: First, the researchers degraded the gratings, by embedding them in visual noise (which looks like the TV snow you see between channels). Sometimes the visual noise contained no gratings, which provided the opportunity for hallucinated gratings to arise. Second, the participants knew that only one grating orientation would be presented, and that the presented orientation would be either 45 degrees or 135 degrees. This knowledge biased participants to develop very specific perceptual expectations.

Previous research had found similar patterns of cortical neural activity for sensory expectations and for actual targets. Other work had noticed equivalent patterns in spontaneous fluctuations in cortical activity. But it was not known whether such fluctuations might predispose the observer to hallucinate.

Pajani and her colleagues found that most of the subjects experienced repeated hallucinations over the course of the experiment–seeing oriented gratings were there were none. Hallucination occurrence was linked to two neural characteristics: a low level of neural activity prior to the visual stimulus (which may or may not have contained an oriented grating), as well as neural activity patterns biased towards the expected grating.

The researchers concluded that hallucinations arose from both imprecise sensory information and cognitive bias. Neither factor, by itself, may suffice to make us see something that is not there..., but in the real world the combination can be perilous, and sometimes fatal. My misperception this morning had zero consequences, and I would have already forgotten all about it, had I not been writing about the topic today. But what if instead of mistaking a notebook for a phone, we mistook a phone for a gun? We may not be able to do much about the quality of the information that arrives to our senses, but we should strive to eliminate cognitive biases that might push our perception in unwelcome directions.