Much of life is predictable. Every day the sun comes up in the east. When we use a toaster bread goes in cool and soft, pops up warm and crispy.
Our brains build models of how the world works, helping us predict what will happen next. Guessing the future helps us survive and the better we are at the game the faster we can react to situations confronting us. Just as important though is our ability to adapt to surprises and integrate new information into our mental world model.
These two steps–registering surprise and learning from it–are two separate mental processes, report a group of neurologists and experimental psychologists from England. The team present the findings in a recent PNAS Early Edition. This is the first time this difference has been detected.
“Surprise and updating are often correlated in experimental paradigms but are, in fact, distinct constructs that can be formally defined,” write the authors. Jill O’Reilly, Urs Shuffelgen and their team found that surprising stimuli triggered behavioral reprogramming in test subjects (seen in slower reaction time) along with activity in the posterior parietal cortex. When the subjects were updating their predictive model a different area, the anterior cingulate cortex, was activated.
Subject’s pupil size also changed based on which mental phase, surprise vs. updating, they were in. In general, pupils expanded during surprise and contracted during updating. Past studies indicate that pupil dilation correlates with the release of noradrenaline. “Theoretically, it has been argued that noradrenaline signals uncertainly about the state of the environment,” write O’Reilly and colleagues, “or that it acts as a kind of ‘reset’ signal for internal models.”
While the dilating effect of surprise is understandable, the update-specific decrease leaves a bit of a mystery. Not only does it clash with current theories of how noradrenaline works, this observation hasn’t been seen in previous experiments. The researchers offer several possible reasons. The learning task tracked in this experiment was unlike previous experiments (saccadic reprogramming instead of probabilistic learning). Also, they were able to look at smaller time scales than previous studies. Finally, “previous studies have not disassociated surprise and updating.”
In some disorders such as Balint syndrome, a person cannot reorient when presented with something unexpected. Reorientation failures are linked to lesions in both the posterior parietal and the anterior cingulate cortex. The authors hope this study will update understanding and potential treatment of related neurological problems.