Downloaded from rspb.royalsocietypublishing.org on July 20, 2010
2 A. E. Welchman et al. Reaction beats intention
(a)
reaction time
execution player 1
time player 2
(c)
reaction time
initiate
react
1!IIIIIIIII
0 200 400 600 800
reaction time
button 2 press button 3 press button 1 press
I________________________________________________I_______________________________________________I_______________________________________________I_______________________________________________I_______________________________________________I________________________________________________I________________
800
lift off
button 1
0 200 400 600
time from movement onset (ms)
Figure 1. (a) An illustration of the button press sequence. Button 1 was referred to as the ‘home key’ and participants initiated a
trial by keeping this button depressed with their right hand. They then moved to the right to hit button 2, then all the way to the
left to hit button 3, before returning to button 1. Buttons were separated laterally by 35 cm (experiment 1) or 15 cm (exper-
iments 2, 3), meaning that arm movement was necessary. (b) An illustration of a single trial competition between two
participants. Players had their own set of three buttons. The movement sequence starts by one player lifting their hand off
button 1, and ends by pressing button 1 again having meanwhile pressed buttons 2 and 3. In this trial player 1 was the initiator
and player 2 the reactor: player 1’s button 1 is lifted up before player 2’s. Player 1 completes the movement sequence first but
player 2 executes the movement faster. Note that this difference in execution times could be spurious: player 2 might simply
make faster movements. Thus, we compared movement times from the same participant—contrasting trials when they were the
initiator with those in which they were the reactor. (c) Distributions of button press times for two representative participants.
Boxplots depict the median, interquartile range and the extreme values; outliers are shown as single points; notches show 95%
CI for the median. The blue boxplots show the distribution of reaction times on ‘reactive’ trials. The green, orange and red
boxplots show the times at which participants depressed buttons 2, 3 and 1, respectively. Separate series are used for reactive
and initiative trials. All times are relative to releasing button 1. As expected for time data, distributions are positively skewed
(Ratcliff 1993). The increasingly broad distributions for buttons 3 and 1 are expected as time is relative to button 1 being
released, so variation is compounded at each subsequent stage.
responses result from observing the movement of the
opponent, and the third tests the importance of a social
context.
2. MATERIAL AND METHODS
(a) Participants
Participants were able-bodied, naive individuals (aged 18-39)
recruited from subject pools in Tubingen and Birmingham.
Ten participants (four males, six females) were used for
experiment 1, 10 participants for experiment 2 (six males,
four females), and 14 for experiment 3 (one male, 13 females).
Further, groups of eight (six males, two females) and 12
(seven males, five females) participants were used in additional
experiments. All gave written informed consent and local
ethics committees approved the experiments.
(b) Equipment
The equipment consisted of two sets of three buttons inter-
faced with a PC through a data acquisition card
(experiment 1) or the parallel port (experiments 2, 3).
Each participant had a set of buttons attached to the table
on which they sat. The buttons were capacitor-based
switches encased in a rigid plastic of 4.5 cm diameter
(Captronic Electronic GmbH), i.e. there were no moving
parts and the buttons did not physically change when
touched. Custom-built electronics converted the button
output to a standard 5 V pulse. Button presses were detected
Proc. R. Soc. B