Summary: Perception is founded on Bayesian principles, even though the brain’s execution of strategies appears to be imperfect. Researchers say the view perception is an imperfect Bayesian inference can provide a middle ground between traditional and anti-Bayesian models.
Human perception is based on mathematically optimal principles, but the brain implements those principles imperfectly, suggests new research by Elina Stengård and Ronald van den Berg of the University of Uppsala, Sweden. They present their findings in PLOS Computational Biology.
The human brain uses imprecise sensory inputs to determine truths about the surrounding environment. Previous research has suggested that human perception is “Bayesian,” meaning that the brain accounts for uncertainty of sensory observations in a mathematically optimal way. However, some of those studies have been criticized mathematically, and other research suggests that the brain is inherently imprecise at the neural level.
To address those concerns, Van den Berg and Stengård presented 30 volunteers with a series of perception tests. These tests involved identifying whether ellipse shapes appearing on a screen were tilted clockwise or counterclockwise from vertical. Different tests incorporated sensory uncertainty in different ways, such as varying degrees of elongation of the ellipse shape, distractions in the form of nearby ellipses, and a short display time of the ellipse on the screen.
The researchers then analyzed their results against a series of different mathematical models. They found that the data is best accounted for by a model that is Bayesian at its core, but also subject to imperfections. This model outperformed both an optimal Bayesian model and all non-Bayesian models that were tested.
“Our results suggest that human perception is blueprinted on optimal strategies, even though the brain’s execution of these strategies seems to be imperfect,” Van den Berg says. “This novel concept provides a theoretical middle ground between the seemingly opposing literature of optimal models and heuristic models.”
Additional research is needed to pinpoint what causes the apparent imperfections in the decision-making process during the ellipse perception tests. Future research could also test whether the imperfect Bayesian model can account for human behavior in other kinds of perception tests, and in higher-level cognitive decision-making tasks.
Funding: RVDB acknowledges support from the Swedish Research Council (Vetenskapsrådet; reg. nr. 2015-00371; http://www.vr.se) and Marie Sklodowska Curie Actions, Cofund (project INCA 600398; https://ec.europa.eu/research). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
About this neuroscience research article
Source: PLOS Media Contacts: Ronald van den Berg – PLOS Image Source: The image is credited to Elina Stengård.
Optimal Bayesian models have been highly successful in describing human performance on perceptual decision-making tasks, such as cue combination and visual search. However, recent studies have argued that these models are often overly flexible and therefore lack explanatory power. Moreover, there are indications that neural computation is inherently imprecise, which makes it implausible that humans would perform optimally on any non-trivial task. Here, we reconsider human performance on a visual-search task by using an approach that constrains model flexibility and tests for computational imperfections. Subjects performed a target detection task in which targets and distractors were tilted ellipses with orientations drawn from Gaussian distributions with different means. We varied the amount of overlap between these distributions to create multiple levels of external uncertainty. We also varied the level of sensory noise, by testing subjects under both short and unlimited display times. On average, empirical performance—measured as d’—fell 18.1% short of optimal performance. We found no evidence that the magnitude of this suboptimality was affected by the level of internal or external uncertainty. The data were well accounted for by a Bayesian model with imperfections in its computations. This “imperfect Bayesian” model convincingly outperformed the “flawless Bayesian” model as well as all ten heuristic models that we tested. These results suggest that perception is founded on Bayesian principles, but with suboptimalities in the implementation of these principles. The view of perception as imperfect Bayesian inference can provide a middle ground between traditional Bayesian and anti-Bayesian views.