The Influence of Posture on Attention

Abstract

Smith et al. (2019) found standing resulted in better performance than sitting in three different cognitive control paradigms: a Stroop task, a task-switching, and a visual search paradigm. Here, we conducted close replications of the authors' three experiments using larger sample sizes than the original work. Our sample sizes had essentially perfect power to detect the key postural effects reported by Smith et al. The results from our experiments revealed that, in contrast to Smith et al., the postural interactions were quite limited in magnitude in addition to being only a fraction of the size of the original effects. Moreover, our results from Experiment 1 are consistent with two recent replications (Caron et al., 2020; Straub et al., 2022), which reported no meaningful influences of posture on the Stroop effect. In all, the current research provides further converging evidence that postural influences on cognition do not appear to be as robust, as was initially reported in prior work.

Keywords: attention; cognitive control; embodiment; posture; standing.

Figure 1. Mean response times and percentage error (presented in parentheses) as a function of posture and congruency from Experiment 1. Error bars represent the 95% confidence intervals calculated according to Loftus and Masson (1994).

Figure 2. Mean response times by congruency, condition, and posture. Error bars represent the 95% confidence intervals calculated according to Cousineau et al. (2021).

Figure 3. Mean percentage error by congruency, condition, and posture. Error bars represent the 95% confidence intervals calculated according to Cousineau et al. (2021).

Figure 4. Mean response times and percentage error (presented in parentheses) as a function of Posture and Congruency from Experiment 3. Error bars represent the 95% confidence intervals calculated according to Loftus and Masson (1994).

Figure 5. Forest plot illustrating the effect sizes of the postural interactions for the RT data reported in three Smith et al. (2019) experiments, as well as the postural effect sizes for three RT data obtained across the three replication experiments reported here. Error bars represent 95% confidence intervals.