S1: Additional Analyses of a Congruency Effect on Response Times

How Personal Standards Perfectionism and Evaluative Concerns Perfectionism Affect the Error Positivity and Post-Error Behaviour with Varying Stimulus Visibility

Drizinsky, Zülch, Gibbons, & Stahl

Supplementary Information

S1: Additional Analyses of a Congruency Effect on Response Times

In order to test the congruency effect on response time (RT) for correct trials, we run a general linear model (GLM) with the factors Congruency (congruent, incongruent) and SOA (100, 66, 33). To reach a sufficient number of trials for this analysis, the factors Response Type and Decision could not be realized. A significant main effect of Congruency was revealed, F(1, 38) = 42.27, p< .001, η2= .52. Responses in incongruent trials were slower (321 ± 11.4ms) than in congruent trials (299 ± 12.1 ms). Furthermore, a significant main effect of SOA was observed, F(1,26) = 10.48, p < .01, η2= .29, showing slower responses for the two longer SOAs(SOA 100: 331 ± 11.7 ms; SOA 66: 322 ± 12.1 ms; no significant difference,p.10) compared to the shorter SOA (SOA 33: 278 ± 13.1 ms; for both comparisons, ps< .001), which could be a result of a fast guessing in the SOA 33 condition. No further significant effects were obtained.

S2: Additional Analyses of Error Negativity

For the sake of completeness, we also performed a GLM with the within-subject factor Decision, Response Type, and SOA for error negativity/correct response negativity. The error negativity was defined as the most negative mean amplitude for the CSD-ERP waveforms within a time window from 0 to 180 ms after the response at FCz.The local maximum of this component was at FCz.

The analyses showed a significant main effect of SOA on the amplitude, F(2, 50) = 6.78, p < .01, η2 = .21. The amplitude of the error negativity/ correct response negativity was smaller in the SOA 33 condition (0.140± 0.022μV/cm2, p< .05) than in the two other SOA conditions (SOA 67: -0.198 ± 0.033μV/cm2; SOA 100: -0.199 ± 0.031μV/cm2), where no significant difference was found (p> .10). In addition, a significant main effect of Response Type was revealed, F(1, 25) = 31.35, p < .001, η2 = .56. The amplitude after errors (0.215 ± 0.032 μV/cm2) was larger compared to correct responses (0.143 ± 0.023μV/cm2). Furthermore, we found a significant SOA x Response Type interaction, F(2,50)=4.43, p < .05, η2 = .15. The amplitude difference between errors and correct responses differed significantly in the two longer SOA conditions (ps< .001 and ps< .01) compared to the short SOA 33 condition, where no significant difference was found (p > .10). Furthermore, a significant interaction between Response Type and Decision was obtained, F(1, 25) = 13.57, p < .01, η2 = .35. If participants committed an error, the amplitude was larger, when they thought that their response was correct (unaware error) than when they thought that they had committed an error (i.e., aware error; p< .05). In correct trials, Decision had no effect on the amplitude of error negativity/correct response negativity. No further significant effects on the amplitude were obtained.

The SOA and Response Type results were in line with the results reported by Gibbons et al. (2011, their 2nd experiment), however, the authors did not assess error awareness. The surprising finding that the error negativity was larger for unaware errors compared to aware errorswas presumably a confounding factor as the amplitudes were averaged across the different SOA conditions, i.e. different causes of a lack of error awareness can be assumed (lack of attention or weak perceptual input). A second GLM including the perfectionism scores as continuous predictors did not reveal a relationship between error negativity/correct response negativity and perfectionism.

S3: Perfectionism-related event-related potentials separated for errors and correct responses (cf. Figure 3 main text)

Supplement Figure 1. Current-source density (CSD) transformed event-related potential waveforms time-locked to the response onset (0 ms) at electrode Cz as a function of ECP for stimulus-onset asynchrony (SOA) 33 for errors (A) and correct responses (B) and ECP and PSP for SOA 100 for errors (C) and correct responses (D).

References

Gibbons, H., Fritzsche, A.-S., Bienert, S., Armbrecht, A.-S., & Stahl, J. (2011). Percept-based and object-based error processing: An experimental dissociation of error-related negativity and error positivity. Clinical Neurophysiology, 122(2), 299−310.doi:10.1016/j.clinph.2010.06.031