Arrow-Cueing: Experimental Manipulation
Table
2 shows the averages and standard deviations of reaction times for all conditions in the arrow- and gaze-cueing blocks. For the arrow-cue trials, the repeated measures GLM with block predictivity (high predictive, non-predictive) and cue validity (valid, invalid) yielded a significant main effect of cue validity,
F(1,86) = 159.569,
p < .001,
η
p
2 = .650 indicating participants were faster on valid trials (515 ms) than on invalid trials (607 ms). In addition, a block predictivity by cue validity interaction was observed,
F(1,86) = 121.724,
p < .001,
η
p
2 = .586, showing larger cueing effects (the difference in RT between a valid and an invalidly cued target) in the high predictive blocks (Δ136 ms) compared to the non-predictive blocks (Δ49 ms). Not surprising, participants rely more on the arrow cues when they are predictive of the location of the target. Further evidence of an increased use of the arrow-cue in predictive blocks, compared to non-predictive blocks could be observed in an increased speeding up for valid trials in the high predictive, compared to the non-predictive blocks (Δ39 ms;
t(86) = 7.367,
p < .001). In addition, a slowed response for invalid cues in the high, compared to the non-predictivity blocks (Δ-49 ms;
t(86) = 8.163,
p < .001) was observed.
Table 2
Reaction times and standard deviation (in ms) for each condition in the experiment separately for the low- and elevated psychopathy groups
| | | Fear | Neutral | |
Low psychopathy group (SRP-SF Tot) | High predictive block | Valid | 512 (74) | 526 (75) | 495 (66) |
| | Invalid | 617 (85) | 610 (80) | 628 (93) |
| Low predictive block | Valid | 524 (82) | 551 (88) | 538 (80) |
| | Invalid | 569 (79) | 573 (88) | 591 (93) |
Elevated psychopathy group (SRP-SF Tot) | High predictive block | Valid | 516 (84) | 520 (79) | 503 (78) |
| | Invalid | 618 (101) | 616 (95) | 634 (109) |
| Low predicitive block | Valid | 557 (91) | 562 (88) | 540 (94) |
| | Invalid | 587 (90) | 593 (89) | 587 (105) |
Gaze-Cueing: Experimental Manipulation
A repeated measures GLM with block predictivity (high predictive, non-predictive), emotion (fear, neutral) and cue validity (valid, invalid) yielded two main effects and three interaction effects. A main effect of cue validity was found, F(1,86) = 151.754, p < .001, η
p
2 = .638, which indicated that participants were faster on valid trials (533 ms), compared to invalid trials (596 ms). A main effect of emotion was found, F(1,86) = 9.672, p = .003, η
p
2 = .101, which showed that participants responded faster when a fearful face was shown (561 ms) than when a neutral face was shown (567 ms).
An interaction effect of block predictivity by cue validity was found, F(1,86) = 78.634, p < .001, η
p
2 = .478. Similar to the arrow cueing experiment, this interaction showed a larger cueing effect in the high predictive blocks (Δ97 ms), compared to the non-predictive blocks (Δ29 ms), suggesting that participants used the gaze cues more efficiently when they were predictive than non-predictive of where the target would appear. Additional post-hoc tests showed that the cueing effects were significant in the predictive blocks, F(1,86) = 146,929, p < .001, η
p
2 = .631, as well as in the non-predictive blocks, F(1,86) = 48.924, p < .001, η
p
2 = .363. The interaction between block predictivity and cue validity shows that the overall predictivity of the cues severely impacts how the cue is used.
An interaction effect of block predictivity by emotion was found (at trend level), F(1,86) = 3.470, p = .066, η
p
2 = .039. In the high predictive block, the average reaction time for fearful faces was 564 ms and 567 ms for neutral faces. In the non-predictive block, the average reaction time was 559 ms for fearful faces and 568 ms for neutral faces, suggesting that the emotional expression plays a more important role in guiding attention when the cues were non-predictive. However, the difference in response times between neutral and fearful faces was only marginally significant in the high predictive blocks, F(1,86) = 3.238, p = .075, η
p
2 = .036, and significant in the non-predictivity blocks, F(1,86) = 8.888, p = .004, η
p
2 = .094.
An interaction effect of emotion by cue validity was found, F(1,86) = 8.440, p = .005, η
p
2 = .089, showing a larger difference in reaction times between neutral and fearful faces when the target was validly cued (Δ10 ms), compared to when the target was invalidly cued (Δ2 ms). Post-hoc testing showed that the difference in reaction time between neutral and fearful faces was only significant for valid cues, F(1,86) = 19.398, p < .001, η
p
2 = .184, but not for invalid cues (F(1,86) = .486, p = .497. From a different perspective, the interaction between emotion and cue validity shows a larger cueing effect for fearful faces (Δ67 ms) compared to neutral faces (Δ59 ms). However, post-hoc tests show that both emotions elicit significant cueing effects (fearful: F(1,86) = 149.636, p < .001, η
p
2 = .635; neutral: F(1,86) = 133.308, p < .001, η
p
2 = .608).
SRP Total
A significant interaction effect of block predictivity by SRP total was observed, F(1,53) = 4.197, p = .045, η
p
2 = .073. A second GLM for the separate groups showed that the block predictivity effect was marginally significant for the low psychopathy group (p = 0.072), and not significant for the elevated psychopathy group (high scoring group; p = 0.289). Numerically, participants in the low psychopathy group showed 18 ms faster responses to validly cued targets in high predictive blocks compared to non-predictive blocks. However, they were 43 ms slower on invalid trials in the high predictive blocks compared to the non-predictive blocks. The slowing down for invalid trials was larger than the speeding up on valid trials for participants in the low psychopathy group, which explains the marginally significant block predictivity effect for this group of participants. The main effect of block predictivity for the elevated psychopathy group was not significant, suggesting that the slowing down on invalid trials is equal to the speeding up on valid trials.
A significant interaction effect of cue validity by emotion by SRP total was observed
F(1,53) = 8.923,
p = .004,
η
p
2 = .144. Post-hoc comparisons confirmed that the interaction between cue validity and emotion is significant for the low psychopathy group
F(1,30) = 28.517,
p < .001,
η
p
2 = .487, but not for the elevated psychopathy group,
F(1,23) = .146,
p = .706. This observation indicates that the elevated psychopathy group shows cueing effects of equal magnitude for either emotion, whereas the low scoring group shows a difference in the magnitude of the cueing effect between the two emotions. Post-hoc t-tests confirmed that the difference between the valid fear trial and valid neutral trial is significantly different in the low psychopathy group,
t(30) = −5.96,
p < .001 but not in the high psychopathy group,
t(23) = −1.043,
p = .308. Table
2 shows that, on valid trials, the low psychopathy group was faster than the elevated psychopathy group when a fearful face was presented. For accuracy, please see (Table
3).
Table 3
Accuracy of the low and elevated SRP-SF groups in the emotional gaze-cueing paradigm. Note that in all conditions both groups perform at ceiling level
Low SRP-SF | 97.7 (93–100%) | 95.8 (85–100%) | 97.7 (91–100%) | 96.1 (78–100%) | 96.5 (85–100%) | 95.6 (84–100%) | 97.6 (88–100%) | 96.5 (83–100%) |
High SRP-SF | 97.1 (93–99%) | 93.0 (73–100%) | 97.2 (89–100%) | 92.3 (76–100%) | 96.1 (76–100%) | 94.1 (68–100%) | 95.3 (76–100%) | 96.0 (84–100%) |