Cardinal rules: Visual orientation perception reflects knowledge of environmental statistics

By Ahna Reza Girshick, Michael S. Landy, Eero P. Simoncelli. Nature Neuroscience, 2011

616 citations recorded by the source corpus.

Humans are good at performing visual tasks, but experimental measurements have revealed substantial biases in the perception of basic visual attributes. An appealing hypothesis is that these biases arise through a process of statistical inference, in which information from noisy measurements is fused with a probabilistic model of the environment. However, such inference is optimal only if the observer's internal model matches the environment. We found this to be the case. We measured performance in an orientation-estimation task and found that orientation judgments were more accurate at cardinal (horizontal and vertical) orientations. Judgments made under conditions of uncertainty were strongly biased toward cardinal orientations. We estimated observers' internal models for orientation and found that they matched the local orientation distribution measured in photographs. In addition, we determined how a neural population could embed probabilistic information responsible for such biases.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/d32ae79a27091626ae7440c68b348fb0682a55a1

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

Citations in the corpus

A Bayesian observer model constrained by efficient coding can explain 'anti-Bayesian' percepts (2015, neutral): “(16) np g © 2 01 5 N at ur e A m er ic a, In c.”
Relation learning in a neurocomputational architecture supports cross-domain transfer (2019, background): “ central reference point—and the RGB encoding of the pixels composing the object. One consequence of this encoding is that the model shows the same bias for the cardinal directions observed in humans (54). This pre-processor is clearly a vast oversimplification of human object perception. However, w”
How Do Expectations Shape Perception? (2018, background): “, horizontal and vertical) lines are more prevalent than oblique ones [21] (Figure 1A), shadows are more likely to appear underneath objects than above them because light usually comes from above [22,23], and objects in the periphery of our visual field often move away from the center of our gaze (c”
Troubles with Bayesianism: An introduction to the psychological immune system (2018, background): “Ironically, it is not entirely implausible that we end up with a picture of the mind in which the faculties of sensation and perception are the home of Bayesian updating (Girshick et al., 2011; Rescorla, 2016) while the workings of cognition bend away from Bayes and towards conceptions of the self.”