Science - USA (2021-11-12)

Thibaultet al.,Science 374 , eabe0874 (2021) 12 November 2021 5 of 14

A

B C

DEF

Fig. 3. Experiments 2 and 3: Cross-domain learning transfer from tool use
to syntactic skills in language.(A) Timeline of training and syntactic pretests
and posttests. Different groups were trained either to use a tool (purple inset)
or the free hand (green inset) to grab and enter key-shaped pegs into grooved holes
or they passively watched videos (gray inset). In Experiment 3, training difficulty
was controlled with a further manual training condition (red inset) with increased
sensorimotor constraints mimicking those present during tool use. (B) Motor
improvement during tool-use training (purple) and free-hand training (green) in
Experiment 2. (C) In Experiment 2, participants with high initial syntactic skills
improved for object relatives in the syntactic task after training with the tool (purple
bars) but not after free-hand training (green bars) or passive video watching (gray

bars). Connected dots across pretest and posttest represent individual data. (D) Motor

improvement during tool-use training (purple), free-hand training (green), and

constrained-hand training (red) in Experiment 3. (E) In Experiment 3, after tool use, the

participants improved for object relatives in the syntactic task (purple bars) but not

after training with the free hand (green bars) or with the hand mimicking similar

sensorimotor constraints as the tool. Red bars indicate significantTraining × Time ×

Sentenceinteraction of the three-way rmANOVA (F(3.1,56.4)= 2.81;P= 0.04;ƞG^2 =

0.03). (F) Effect size of the improvements in Experiments 2 and 3. Two-wayTraining ×

ExperimentrmANOVA: There was a significant main effect ofTraining(F(1,60)= 5.37;

P= 0.02;ƞG^2 = 0.082), a nonsignificant effect ofExperiment(F(1,60)= 2.8;P= 0.11),

and no interaction (F(1,60)= 0.51;P= 0.58). ***P<0.001;**P< 0.01; *P< 0.05.

RESEARCH | RESEARCH ARTICLE