Large inter-individual and intra-individual variability in the effect of perceptual load

Abstract

This study examined whether the recurrent difficulty to replicate results obtained with paradigms measuring distractor processing as a function of perceptual load is due to individual differences. We first reanalyzed, at the individual level, the data of eight previously reported experiments. These reanalyses revealed substantial inter-individual differences, with particularly low percentage of participants whose performance matched the load theory's predictions (i.e., larger distractor interference with low than high levels of load). Moreover, frequently the results were opposite to the theory's predictions-larger interference in the high than low load condition; and often a reversed compatibility effect emerged-better performance in the incompatible than neutral condition. Subsequently, seven observers participated in five identical experimental sessions. If the observed inter-individual differences are due to some stable trait or perceptual capacity, similar results should have emerged in all sessions of a given participant. However, all seven participants showed large between-sessions variations with similar patterns to those found between participants. These findings question the theoretical foundation implemented with these paradigms, as none of the theories suggested thus far can account for such inter- and intra-individual differences. Thus, these paradigms should be used with caution until further research will provide better understanding of what they actually measure.

Fig 1

Examples of the stimuli employed in all the experiments: (a) Low load condition; (b) High load condition.

Fig 2

Distractor interference in the high vs. low load condition for the aggregated data (averaged across participants; in red) and for each participant (in blue) in Experiment 1: (a) RT; (b) accuracy. Error bars represent 95% confidence intervals (bootstrapped for the individual data points). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 3

Distractor interference in the high vs. low load condition for the aggregated data (averaged across participants; in red) and for each participant (in blue): (a) Experiment 2a—RT; (b) Experiment 2b—RT; (c) Experiment 2a—accuracy; (d) Experiment 2b—accuracy. Error bars represent 95% confidence intervals (bootstrapped for the individual data points). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 4

Distractor interference in the high vs. low load condition for the aggregated data (averaged across participants; in red) and for each participant (in blue): (a) Experiment 3—RT; (b) Experiment 4—RT; (c) Experiment 3—accuracy; (d) Experiment 4—accuracy. Error bars represent 95% confidence intervals (bootstrapped for the individual data points). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 5

Distractor interference in the high vs. low load condition for the aggregated data (averaged across participants; in red) and for each participant (in blue) in Experiment 2: (a) 100 ms condition—RT; (b) 150 ms condition—RT; (c) 100 ms condition—accuracy; (d) 150 ms condition—accuracy. Error bars represent 95% confidence intervals (bootstrapped for the individual data points). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 6

Distractor interference in the high vs. low load condition for the aggregated data (averaged across participants; in red) and for each participant (in blue) in Experiment 3: (a) 100 ms condition—RT; (b) 150 ms condition—RT; (c) 100 ms condition—accuracy; (d) 150 ms condition—accuracy. Error bars represent 95% confidence intervals (bootstrapped for the individual data points). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 7

Distractor interference in the high vs. low load condition for the aggregated data (averaged across participants; in red) and for each participant (in blue) in Experiment 4: (a) 100 ms condition—RT; (b) 150 ms condition—RT; (c) 100 ms condition—accuracy; (d) 150 ms condition—accuracy. Error bars represent 95% confidence intervals (bootstrapped for the individual data points). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 8. Distractor interference (incompatible vs. neutral) and distractor facilitation (compatible vs. neutral) as a function of load in the first session of the current experiment.

(a) RT; (b) accuracy. '*' indicates significant distractor effect.

Fig 9

Distractor interference in the high vs. low load condition in the first experimental session of the current experiment: (a) RT; (b) accuracy. Aggregated data is in red. Data points for the 13 participants who only participated in this first session are in blue. The seven participants who participated in additional experimental sessions are marked with other colors (see text for details). Error bars represent 95% confidence intervals (bootstrapped). Note that due to the large axes scale, individual confidence intervals are smaller than the symbol of their corresponding data point.

Fig 10. RT distractor interference (incompatible vs. neutral) as a function of load condition and session for each of the seven participants.

Error bars represent 95% confidence intervals (bootstrapped).

Fig 11. Accuracy distractor interference (incompatible vs. neutral) as a function of load condition and session for each of the seven participants.

Error bars represent 95% confidence intervals (bootstrapped).