Open hands, large numbers: manual gestures influence random number generation

Abstract

Several studies suggest that numerical cognition interacts with spatial cognition. Here, we explored spatial-numerical associations through the lens of manual gestures. We asked English and Italian participants to generate 'random' sequences of numbers while simultaneously moving the hands outwards, away from the torso, or inwards, towards the mid center of the body. These manual gestures were modeled after gestures that are common in naturally occurring numerical discourse, such as when people talk about "huge numbers" or "tiny numbers." Results showed that in both participant groups, outwards movements coincided with relatively larger numbers compared to inwards movements, for which generated numbers were smaller. This effect was small in magnitude. We also explored individual differences and found that self-reported numeracy as well as levels of gesture production and perception modulated the effect of our gesture manipulation very little, if at all.

Fig. 1

Sample a outwards and b inwards gestures. Drawing by Alexandra Lorson

Fig. 2

Pair plot showing average number magnitudes for inwards trials (yellow) and outwards trials (purple); each line represents one participant, with line color indicating whether each participant’s effect was congruent (black) or incongruent (gray) with our hypothesis

Fig. 3

Posterior distributions for the three fixed effects in the main model (mixed beta-binomial regression on absolute magnitudes) that includes the gesture by language interaction

Fig. 4

Posterior distributions for the three fixed effects in the logistic regression model (relative differences) that includes the gesture by language interaction

Fig. 5

Participant average difference score (up = higher numbers for outwards trials) plotted against a) the ‘Brief Assessment of Gesture’ (BAG) scale, and b) the ‘Subjective Numeracy Scale’ (SNS); point colors indicate English (light grey) and Italian (dark blue)

Fig. 6

Dot plot showing the percentage difference (outwards minus inwards) for each of the numbers 1–30; vertical bars highlight whether changes are above or below the 50% midline (no difference between outwards and inwards). Y-axis range is constrained to 40–60% to make differences between numbers visible