A Quantized Representation of Probability in the Brain
James Tee, Desmond P. Taylor

TL;DR
This paper presents evidence that the brain encodes probabilities in a quantized, discrete manner rather than as continuous values, revealing a fundamental limit in neural probabilistic representation.
Contribution
It introduces a novel quantized probability distortion function and provides empirical evidence supporting the brain's use of 4-bit quantized probability representations.
Findings
78% of participants fit quantized models better than continuous ones
Strong evidence for 4-bit probability quantization in neural representation
Reveals a fundamental precision limit in brain's probabilistic encoding
Abstract
Conventional and current wisdom assumes that the brain represents probability as a continuous number to many decimal places. This assumption seems implausible given finite and scarce resources in the brain. Quantization is an information encoding process whereby a continuous quantity is systematically divided into a finite number of possible categories. Rounding is a simple example of quantization. We apply this information theoretic concept to develop a novel quantized (i.e., discrete) probability distortion function. We develop three conjunction probability gambling tasks to look for evidence of quantized probability representations in the brain. We hypothesize that certain ranges of probability will be lumped together in the same indifferent category if a quantized representation exists. For example, two distinct probabilities such as 0.57 and 0.585 may be treated indifferently. Our…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
