Does the brain code 'how many' with two channels or many?
After staring at 50-dot arrays, people saw smaller arrays as even fewer and larger arrays as even more numerous, a pattern that fits many number-tuned channels rather than two opposing ones.
Source
Visual adaptation reveals multichannel coding for numerosity
Study at a glance
- Design
- Human experiment — Within-subject control-then-adaptation dot-array comparison task, with a between-subject small (10-50) versus large (50-250) numerosity range; adaptor fixed at 50 dots.
- N
- N=8 · 8 adults, 4 per stimulus-range condition, each completing 240 control and 240 adaptation trials; a footnoted extra experiment used a 50-130 range but its N is not stated.
- Population
- University students at Emory University with normal vision
- Outcome
- Proportion of trials judging the adapted-location test array as more numerous (adaptation minus control)
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Eight students compared the numerosity of two briefly flashed dot arrays, one in a location that had been repeatedly exposed to 50-dot adaptor arrays and one in an unadapted location. Half the participants saw arrays from 10 to 50 dots and half from 50 to 250 dots. Everyone did a control block without adaptation first, then the adaptation block, and the researchers compared how often the adapted array was judged more numerous.
What they found
Without adaptation, accuracy was well above chance and fell as the two numbers got closer in ratio, as Weber's law predicts. After adapting to 50 dots, arrays smaller than 50 were underestimated (d = 0.72) and arrays larger than 50 were overestimated (d = 0.25), and the two ranges differed clearly from each other (d = 0.95). This bidirectional, repulsive aftereffect is what a multichannel code predicts, whereas an opponent two-channel code predicts either no effect or a shift in one direction only.
The limits
What it doesn't show
The sample was tiny, with only four people per range, and range was manipulated between participants, so individual differences could contribute. Dot size and field area were fixed, meaning total dot area and density rose with number, so the aftereffects might partly reflect adaptation to those non-numerical cues. The control task always came first, so practice or fatigue could be confounded with adaptation, and the study infers neural coding from behaviour without measuring the brain.
Key terms
- Perceptual adaptation
- A temporary shift in perception after prolonged exposure to a stimulus, such as seeing later arrays as more or less numerous.
- Opponent channel coding
- A scheme where a dimension is represented by two channels tuned to opposite extremes, so adaptation shifts all percepts in one direction.
- Multichannel coding
- A scheme with many channels each tuned to a preferred value, so adapting one value pushes nearby percepts away from it in both directions.
- Weber's law
- The principle that discrimination depends on the ratio between two magnitudes, so closer ratios are harder to tell apart.
- Approximate number system
- A non-verbal system for rapidly estimating quantities without counting.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What was the adaptor numerosity in this study?
Common questions
Why adapt to 50 dots rather than a very small or very large number?
Only a middle value lets you test numbers both below and above the adaptor, which is needed to tell a one-direction shift from a two-direction repulsion.
Could the result reflect texture or density perception instead of number?
The authors checked whether smaller and larger values within the 50-250 range behaved differently and found no difference, but because area and density covaried with number they cannot fully rule out non-numerical cues.
How does this connect to brain research?
Multichannel coding resembles 'labeled-line' neurons in monkey parietal cortex that each prefer a particular number, though this study did not record brain activity.
More on Numerical cognition