B.2.6.2. Results

In the auditory task, participants omitted 9 % of the target, in average. False alarms were exceptional. RT outliers were discarded in the same way as in Experiment 1. Owing to RT outliers or errors, 3,8 % of the trials were discarded. An analysis of variance was carried out on the remaining correct RTs, with task (simple task, dual task), condition (base, distractor) and display size (3, 6 items) as within-subject factors. Again, the main task was achieved faster in the single task than in the dual task condition (505 ms vs. 531 ms ; F(1,27) = 7.3; SEM = 38,325; p < .012). A Student t-test confirmed that this difference was also significant when considering only the no-distractor condition (t(27)=2.77; p < .01). The presence of the distractor induced an increase of the RTs (532 ms vs. 504 ms; F(1,27) = 31.3; SEM = 45,429; p < .001). On the contrary, the effect of size failed to reach significance (516 ms for size‑3 vs. 520 ms for size‑6; F(1,27) = 2.8, p > .10). The condition X display size interaction was significant (F(1,27) = 7.3; SEM = 3,395; p < .012): again, AC was larger in the size-3 (34 ms) than in the size-6 condition (23 ms), consistent with the perceptual load hypothesis. The two one-way interactions involving the task did not reach significance: neither task X condition (F(1,27) = 2.3; SEM = 2,302; p = .14), nor task X display size (F(1,27) = 0.7; SEM = 193; p > .41 ).

However, the task X condition X display size three-way interaction was marginally significant (F(1,27) = 4.1; SEM = 4,116; p < .053). Consequently, we conducted two ANOVAs on correct RTs, one for each level (3 and 6) of the factor Size, with task (simple task, dual task) and condition (base vs distractor) as within-subject factors. When three items were presented, the effects of task (F(1,27) = 6.2; SEM = 16,539; p < .020) and condition (F(1,27) = 31.6; SEM = 36,830; p < .001) were significant. More crucially, the AC increased from single task to dual task (F(1,27) = 4.9; SEM = 2,770; p < .035), consistently with the cognitive load hypothesis. When six items were presented, the effects of task (F(1,27) = 7.6; SEM = 21,980; p < .011) and condition (F(1,27) = 15.9; SEM = 11,994; p < .001) were significant. However, contrary to the size-3 condition, AC did not vary significantly with task (F(1,27) = 0.6; SEM = 27; p > .80). Again, this seemed consistent with the perceptual load hypothesis, which predicted that AC would be weaker in the high perceptual load condition (6 items), and thus, probably, less prone to be modulated by the cognitive load.

The present experiments also aimed at supporting that attentional orienting and resistance to interference were linked. To that end, we computed a correlation between two indices: a) the effect of WM load on simple attentional orienting and b) the effect of WM load on resistance to interference. For a), we subtracted mean RTs of the no-onset condition in the single task from those in the dual task. For b), we subtracted the AC effects (i.e. the slowing down due to the presence of the distractor) in the single task from those in the dual task. Then, we computed the r² coefficient of correlation, and tested its significance with a Student t-test. To increase the power of the test, we computed the correlation for the results of both experiments together. The coefficient of correlation reached the significant value of r² = 0.26 (t(46)=1.84; p < .05).

Finally, to assess the discrepancies between the two experiments, a general ANOVA was conducted, with task (single task, dual task), condition (distractor, no-distractor) and display size (3 items, 6 items) as within-subject factors and experiment (Exp. 1, Exp. 2) as a between factor. The complete four-way interaction was not significant (F(1,46) = 0.9; SEM = 338; p > .34), suggesting that the the three-way interaction between task, condition and display size did not differ significantly from one experiment to the other. On the contrary, consistently with the hypotheses, the two-way interaction between condition and task was significant (F(1,46) = 9.3; SEM = 7,117; p < .004), while the interaction between condition and display size was close to be (F(1,46) = 3.6; SEM = 2,026; p < .065).