Skip to content
PaperFren

Does how you ask change how well people estimate dot numbers?

Open paper intelligence

People badly underestimated how many dots they saw, yet judged the ratio between two dot sets almost perfectly.

Source

Task Constraints Affect Mapping From Approximate Number System Estimates to Symbolic Numbers

Chesney DL, Matthews PG · Frontiers in psychology · 2018

doi.org/10.3389/fpsyg.2018.01801Read the full paper ↗9 citationscc by

Study at a glance

Design
Human experiment — Within-subject comparison of three estimation tasks (free, number-line, ratio) using briefly flashed dot arrays of 20 to 300 dots with area and dot-size controls
N
N=31 · 31 undergraduates, each completing all three tasks (51 analysed trials per task)
Population
Undergraduates at a selective private US university
Outcome
Median numerical estimates fitted to linear, logarithmic, power and cyclical power models; response variability

Structured fields used in claim comparison tables when every cited study has a complete layer.

What they did

Undergraduates saw arrays of between 20 and 300 dots flashed too briefly to count. In one block they estimated the ratio of an array to a 300-dot array, in another they marked its position on a line anchored by 1 dot and 300 dots, and in a third they simply typed how many dots there were. Median answers for each array size were fitted to several mathematical models.

What they found

Free estimates were linear but only about one third of the true value, so a 300-dot array was called roughly 100. Number-line estimates were intermediate, too high for small arrays and too low for large ones. Ratio estimates were close to exact, with a slope near 1. Only free estimation showed variability that grew with set size (scalar variability).

The limits

What it doesn't show

The sample was small and drawn from one selective university, and the task order was fixed (ratio, then number line, then free), so order or fatigue effects cannot be ruled out. The idea that people 'auto-scale' to 100, and the broader claim that the number sense is relational, are the authors' speculative interpretations rather than tested mechanisms. The number-line results did not match the predicted cyclical pattern at the endpoints, possibly because of memory demands from the brief display, and no arrays below 20 were tested.

Key terms

Approximate number system (ANS)
An intuitive system for sensing roughly how many items there are without counting.
Free estimation
Giving a number for how many items were seen, with no upper bound supplied.
Number-line estimation
Marking where a quantity falls on a line whose two ends are fixed reference amounts.
Scalar variability
The pattern where the spread of estimates grows in proportion to the size of the quantity.
Subitizing
Instant, accurate recognition of very small numbers of items, up to about four.

Flashcards

1 / 9

0 of 9 answers reviewed

Research intelligence for this paper

See its role on concept claims, tensions it is part of, placement history, and related discoveries.

Open paper intelligence

Quiz yourself

1 / 5

In free estimation, participants' answers were roughly what fraction of the true number?

Common questions

Why were the dots flashed so briefly?

To stop people from counting, so estimates had to come from the approximate number system rather than exact enumeration.

Why does it matter that ratio estimates were accurate?

It suggests the number sense may be better at perceiving relations between quantities than absolute amounts, so children might link symbols to quantities through ratios rather than a one-to-one map.

Were free estimates random?

No. They were very consistent, just scaled down to about a third of the true number.

More on Numerical cognition