Prospective Memory Functioning

In order to determine statistical significance of the hypotheses-relevant effects—

stress effects and a potential demand dependence of stress effects on PM performance, PM-monitoring costs, aftereffects of completed intentions and commission-error risk—we used a Bonferroni-corrected alpha level of .002 per test (.05/24). Specifically, we corrected for overall 24 tests of group differences in the hypotheses-relevant dependent variables. Note, that p values for tests at the Bonferroni-corrected alpha level are denoted as p*.

Nonfocal condition. For PM performance, we conducted mixed ANOVAs involving the factors group (stress vs. no-stress) and prospective load (low vs. high) on RTs and error rates on PM trials and false-alarm PM responses during ongoing-task performance. The effects of interest were a main effect of group and a Group × Prospective load interaction.

To examine monitoring costs, we conducted 2 Block (PM vs. test block) × 2 Prospective load (low vs. high) × 2 Group (stress vs. no-stress) ANOVAs on ongoing-task RTs and error rates.

Here our main focus lied on the potential interactions of Group × Block and Group × Block ×

0

Prospective load. For the analysis of aftereffects of completed intentions, we compared performance in PMREPEATED and ongoing-task trials during test blocks in 2 Prospective load (low vs. high) × 2 Trial type (PMREPEATED vs. ongoing task) × 2 Group (stress vs. no-stress) ANOVAs (see Figure 10). Note that aftereffect analyses in the nonfocal condition were conducted exclusively for test blocks following nonfocal PM blocks, since only these

contained PMREPEATED trials. In the aftereffect analyses, the effects of interest were a Group × Trial type and a Group × Trial type × load interaction.

PM performance. RTs on nonfocal PM cues did not differ between stress (1197 ms) and no-stress group (1240 ms), F(1, 78) = .845, p* = .361, ηp2 = .01. PM responses were slower under high (1256 ms) compared to low prospective load (1181 ms), F(1, 78) = 8.64, p = .004, ηp2 = .10. Although this PM-response slowing under high compared to low prospective load was less pronounced in the stress group (43 ms, drm = 0.19) compared to the no-stress group (108 ms, drm = 0.42), this difference failed statistical significance, F(1, 78) = 1.62, p* = .207, ηp2 = .02. Treatment had no statistically significant effect on PM errors (stress: 25.8%; no-stress: 20.8%), F(1, 78) =1.75, p* = .190, ηp2 = .02, or false-alarm rates (stress: 0.5%; no-stress: 0.4%), F(1, 78) = .58, p* = .449, ηp2 = .01. No further effects were significant, all Fs ≤ 2.71, ps ≥ .104, ηp2s ≤ .03.

Monitoring costs. Monitoring costs were reflected in substantially slower ongoing-task performance during PM blocks (1023 ms) compared to test blocks (704 ms), F(1, 78) = 419.96, p < .001, ηp2 = .84. Most importantly, these monitoring costs were markedly lower in the stress group (269 ms, drm = 1.42) than in the no-stress group (371 ms, drm = 1.80), F(1, 78)

= 10.66, p* = .002, ηp2 = .12 (see Table 6). Further, monitoring costs were more pronounced under high (342 ms, drm = 1.59) compared to low prospective load (297 ms, drm = 1.43), F(1, 78) = 13.26, p < .001, ηp2 = .15. This effect, however, was only numerically smaller in the stress (27 ms, drm = 0.19) than in the no-stress group (64 ms, drm = 0.37), F(1, 78) = 2.30, p* = .134, ηp2 = .03. Finally, ongoing-task responses were faster in the stress (818 ms) compared to the no-stress group (909 ms), F(1, 78) = 7.31, p = .008, ηp2 = .09, and participants

responded slower under high (874 ms) than under low prospective load (853 ms), F(1, 78) = 10.17, p = .002, ηp2 = .12. This load-dependent slowing did not differ between stress (16 ms, drm = 0.11) and no-stress group (27 ms, drm = 0.17), F(1, 78) = .64, p = .427, ηp2 = .01.

Error rates mirrored the RT data and were increased during PM blocks (4.7%)

compared to test blocks (3.8%), indicating the presence of monitoring costs during PM task performance, F(1, 78) = 17.26, p < .001, ηp2 = .18. While these costs were again nominally

smaller in the stress (0.3%, drm = 0.09) than in the no-stress group (1.6%, drm = 0.54), this difference did not reach significance at the Bonferroni-corrected alpha level, F(1, 78) = 8.61, p* = .004, ηp2 = .10. Monitoring costs were increased during high (1.6%, drm = 0.48) compared to low prospective load (0.3%, drm = 0.09), F(1, 78) = 10.68, p = .002, ηp2 = .12. This increase in monitoring costs, however, was similar in the stress (1.2%, drm = 0.42) and in the no-stress group (1.5%, drm = 0.57), F(1, 78) = .12, p* = .726, ηp2 = .00. No further hypotheses-relevant effects were significant, all Fs ≤ 1.20, ps ≥ .277, ηp2s ≤ .02.

Aftereffects of completed intention. Participants’ responses were slower in PMREPEATED

(757 ms) compared to ongoing-task trials (695 ms), F(1, 78) = 29.10, p < .001, ηp2 = .27, which indicated the presence of aftereffects of completed intentions (see Table 7). Most

importantly, aftereffects did not differ between groups (stress: 50 ms, drm = 0.35; no-stress:

73 ms, drm = 0.38), F(1, 78) = .997, p* = .321, ηp2 = .01. We found no further statistically significant effects on RTs, all Fs ≤ 3.10, ps ≥ .082, ηp2s ≤ .04, or any effects in error rates, all Fs ≤ 1.97, ps ≥ .164, ηp2s ≤ .03. Since only one participant made a single commission error in the nonfocal condition (stress group, low prospective load) we did not conduct a

commission-error analysis.

Figure 10. Results of the nonfocal condition. Mean response times (RT) and error rates for (A) PM performance, (B) monitoring costs, and (C) aftereffects of completed intentions as a function of prospective load (low vs.

high), group (stress vs. no-stress), block (only for monitoring costs: PM vs. test block), and trial type (only for aftereffects of completed intentions: PMREPEATED vs. oddball). For aftereffects of completed intentions, data were obtained only from test blocks that followed PM blocks, since only these contained PMREPEATED trials. Error bars represent standard errors of the mean.

4000

Low Prospective load High Low Prospective load High

Low High

Prospective load Monitoring costs Aftereffects of completed intentions

PM performance

A B C

Stress No-stress

Focal condition. Analyses of the focal condition were similar to the nonfocal condition but did not include the within-subject factor prospective load. For aftereffect analyses, we contrasted performance on PMREPEATED and oddball trials (see Figure 11). As in the analyses for the nonfocal condition, we determined statistical significance for the effects of interest at a Bonferroni-corrected alpha level of .002 per test (.05/24).

PM performance. Analyses did not reveal a significant effect of stress on PM performance, as RTs were similar in the stress (786 ms) and the no-stress group (786 ms), F(1, 78) < .001, p* = .990, ηp2 < .001, and error rates did not differ between groups (stress:

11.0%; no-stress: 8.3%), F(1, 78) =1.16, p* = .285, ηp2 = .02. Participants committed more false alarms during oddball (2.7%) than during standard trials (0.1%), F(1, 78) = 29.23, p <

.001, ηp2 = .27. All further effects were not significant, Fs ≤ 1.42, ps ≥ .238, ηp2s ≤ .02.

Monitoring costs. Performance decrements during PM blocks (760 ms) compared to test blocks (668 ms) indicated monitoring costs, F(1, 78) = 181.75, p < .001, ηp2 = .70.

However, in contrast to the nonfocal condition, monitoring costs did not differ statistically between stress (87 ms, drm = 0.65) and no-stress group (98 ms, drm = 0.66), F(1, 78) = .67, p* = .415, ηp2 = .01. Analyses also revealed a small to moderate, but nonsignificant, trend for faster ongoing-task responses in the stress (690 ms) compared to the no-stress group (739 ms), F(1, 78) = 3.64, p = .060, ηp2 = .05 (see Table 6). Error rates did not differ between groups (stress: 3.7%; no-stress: 2.8%), F(1, 78) = 3.03, p = .086, ηp2 = .04, but were lower during PM blocks (3.0%) than during test blocks (3.5%), F(1, 78) = 7.78, p = .007, ηp2 = .10. This block effect, however, did not differ between stress (−0.7%, drm = −0.24) and no-stress group (−0.5%, drm = −0.21), F(1, 78) = .19, p* = .667, ηp2 = .002.

Aftereffects of completed intentions. Slower responses towards PMREPEATED (821 ms) than towards oddball trials (732 ms) indicated the presence of aftereffects of completed intentions, F(1, 78) = 69.94, p < .001, ηp2 = .47 (see Table 7). Aftereffects did not differ between stress (72 ms, drm = 0.51) and no-stress group¹⁰ (106 ms, drm = 0.65), F(1, 78) = 2.44, p* = .122, ηp2 = .03. Again, the stress group responded numerically faster (754 ms) than the no stress group (800 ms), which did not reach significance, F(1, 78) = 2.34, p = .130, ηp2 = .03.

There were no significant effects on error rates, all Fs ≤ 1.42, ps ≥ .238, ηp2s ≤ .02.

Similar to previous research on commission errors (e.g., Anderson & Einstein, 2017;

Scullin et al., 2012), we analyzed the proportion of participants who made at least one

10 This result remained unchanged when excluding participants who made commission errors.

commission error on standard, PMREPEATED, or oddball trials in test blocks after treatment¹¹. While on a descriptive level, commission errors were more frequently made by participants in the stress (11 participants) than in the no-stress group (3 participants), this difference did not reach the Bonferroni-corrected alpha level of significance, χ²(1) = 5.54, p* = .019, odds ratio = 4.63.

Figure 11. Results of the focal condition. Mean response times (RT) and error rates for (A) PM performance, (B) monitoring costs, and (C) aftereffects of completed intentions as a function of group (stress vs. no-stress), block (only for monitoring costs: PM block vs. test block), and trial type (only for aftereffects of completed intentions:

PMREPEATED vs. oddball). Error bars represent standard errors of the mean.

Comparison of focal and nonfocal condition. In order to determine whether stress differentially affected monitoring costs in the focal compared to the nonfocal condition, we conducted mixed ANOVAs involving the factors group (stress vs. no-stress) and PM condition (focal vs. nonfocal) on monitoring costs in RTs and error rates (averaged across low and high prospective load in the nonfocal condition). For RTs, monitoring costs were increased in the nonfocal (320 ms) compared to the focal condition (92 ms), F(1, 78) = 280.94, p < .001, ηp2 = .78. Most importantly, a significant Group × PM condition interaction, F(1, 78) = 11.15, p* = .001, ηp2 = .13, revealed a more pronounced reduction of monitoring costs under stress in the nonfocal (102 ms) than in the focal condition (11 ms). For error rates, we also found larger monitoring costs in the nonfocal (0.9%) compared to the focal condition (−0.6%), F(1, 78) = 22.28, p < .001, ηp2 = .13. However, the Block × Group × PM condition interaction did not reach significance, F(1, 78) = 3.15, p* = .080, ηp2 = .04 (nonfocal: 1.3%, focal: 0.2%).

11Commission errors occurred across all focal cycles after treatment and almost exclusively during the first encounter of a PMREPEATED cue within a block, except for one participant who made a commission error on the first deviant trial that was presented in the block (i.e., oddball trial).

4000

PM performance Monitoring costs Aftereffects of completed intentions

A B C

Stress No-stress

Table 7

Assessment of aftereffects of completed intentions. Mean RTs, error rates and false-alarm PM responses during test blocks by trial type (ongoing task, PMREPEATED, oddball) (standard deviations in parentheses).

Stress group No-stress group

Trial type RT (ms) Errors (%) False PM (%) RT (ms) Errors (%) False PM (%) Nonfocal condition, low prospective load^a

Test block Ongoing task 673 (182) 4.5 (3.6) 0 723 (141) 3.1 (2.7) 0

PMREPEATED 726 (182) 3.1 (8.4) 0 793 (244) 2.5 (7.6) 0

Aftereffects^b 53 (120) −1.4 (8.6) - 63 (140) −0.6 (7.0) -

Nonfocal condition, high prospective load^a

Test block Ongoing task 674 (100) 4.2 (4.2) 0 709 (118) 3.0 (3.5) 0

aData for the nonfocal condition was obtained from test blocks that followed PM blocks only. Since only these contained PMREPEATED trials.

bAftereffects represent performance differences between PMREPEATED and ongoing-task (nonfocal condition) or oddball trials (focal condition).

Table 6

Assessment of monitoring costs. PM and ongoing-task performance in mean RTs, error rates, and false-alarm PM responses during PM and test blocks by trial type (ongoing task, PM, oddball) (standard deviations in parentheses).

Stress group No-stress group

Trial type RT (ms) Errors (%) False PM (%) RT (ms) Errors (%) False PM (%) Nonfocal condition, low prospective load

PM block PM 1175 (233) 27.2 (20.8) - 1186 (219) 20.0 (16.9) -

Ongoing task 937 (191) 4.3 (2.7) 0.4(0.6) 1065 (211) 4.3 (3.2) 0.4 (0.7)

Test block Ongoing task 682 (117) 4.7 (3.2) 0 726 (127) 3.4 (2.2) 0

Monitoring costs^a 256 (119) −0.3 (2.9) - 338 (141) 0.9 (2.4) -

Nonfocal condition, high prospective load

PM block PM 1218 (228) 24.4 (21.0) - 1294 (276) 21.6 (20.8) -

aMonitoring costs represent differences in ongoing-task performance between PM blocks and test blocks.