The second ANOVA, relative to distracter effects, showed instead an overall significant effect of reward bias (F 1,30) = 5.969, p<0.03, η2p = 0.166), indicating a small but reliable global disadvantage in responding on trials with a highly rewarded distracter (749.7 ms, 80% high/20% low vs. 739.7 ms, 20% high/80% low).
In order to investigate whether the slight advantage in responding on same trials in the first block was reliable, a further ANOVA was conducted on this block, but this also was nonsignificant, F < 1.
Participants were slower to respond on trials in which threat items were absent compared to present, just as expected in a self-terminating search task.
E.g., if a person was on average 10 ms slower to respond on trials with angry than neutral cues on the central cue task (which reflects a 10 ms slowing effect of threat on RT, independent of location-based spatial cueing effects), this value was subtracted from their mean RT from trials with angry face cues in each condition (valid and invalid) of the spatial cueing task.
ANOVA with block and trial type revealed a significant interaction, F 1,15) = 4.64, p = .048, and responding on different trials was higher than that on same trials on block 1, F 1,15) = 6.34, p = .024.024
In Test 1 the rate of responding on same trials was 10.65 and 10.06 rpm, and on different trials 10.28 and 9.72 rpm, for blocks 1 and 2, respectively; the corresponding means for Test 2 were 7.47 and 7.30, and 7.45 and 7.27 rpm.
Here responding on different trials again appeared greater on block 1 (bottom panel of Fig. 7).
But why should summation of two Pavlovian CSs produce more responding on different trials?
If the occasion setter is not present, then Pavlovian processes will predominate – which in this procedure would produce the opposite pattern, more responding on different trials.
However, in contrast to the hierarchical account, the pseudo-configural theory cannot accommodate the opposite result, greater responding on different trials.
The rates of responding to the transfer CS alone were 4.59 and 3.03 rpm in blocks 1 and 2, respectively, and this was lower than responding on compound trials, F 1,15) = 15.90, p = .001.001
