This effect was further qualified by a near-significant condition-by-time-by-group (F 1,62 = 3.911, p = 0.052, ηp2 = 0.059) as well as a significant condition-by-time- by-valence- by-group (F 1,62 = 4.040, p = 0.049, ηp2 = 0.061) interaction.
In HC, the opposite pattern emerged, with marginally significant negative correlations between learning from negative feedback and ∆FRN in the first ( r = −0.347 , p = 0.060) and the second half ( r = −0.350 , p = 0.058) of the observation condition, but not the control condition ( r = −0.155 , p = 0.412 and r = −0.275 , p = 0.142, respectively).
For positive learning, the correlation with ∆FRN amplitudes in the second half of the experiment approached significance in SAD ( r = −0.297 , p = 0.088), but no further effects were found in either SAD (all p s > 0.288) or HC (all p s > 0.168). ∆P3 In SAD, the ∆P3 in the second half of the observation condition ( r = 0.541 , p = 0.002) showed a significant positive relationship with learning from positive feedback, and ∆P3 amplitudes in the first half of the control condition also tended to correlate with positive learning ( r = 0.312 , p = 0.094).
All other main and interaction effects failed to reach significance (all p s > 0.125).