Abstract
In the Attentional Blink (AB) effect, the identification of Target 2 (T2) in a stream is impaired when presented 200-500ms after Target 1 (T1). Recent studies suggest that higher-order relationships between targets (e.g., semantically associated pairs) modulate the AB effect (Potter et al., 2005; Ferlazzo et al., 2007). In this study, we ask whether an implicitly learned temporal association among targets modulates the AB effect. To test this question, we manipulated the temporal association between targets through statistical learning before an AB task. The experiment consisted of two phases: Training and Letter Detection (AB task). During training, all participants saw a stream of 1000 letters while performing a one-back task. Participants were randomly assigned to view a stream of letters in random order (Random group), or a letter stream with an embedded triplet structure (Statistical group) consisting of 12 random letters organized into four triplets with non-uniform transitional probabilities within and across triplets (e.g., A-K-T, F-J-N). After training, participants performed a letter detection task by reporting two letter targets among a stream of rapidly presented digits (ISI = 100ms). Critically, we manipulated, within subject, whether targets were drawn from the same (within-) or different (across-) triplets. For each trial type, T2 was presented downstream from T1 with lag of 2 or 8 intervening digits. Both groups showed a clear AB effect (Lag8 identification > Lag2). However, only the Statistical group showed a modulation of the AB effect for within-triplet trials compared to across-triplet trials: Lag2 accuracy for within-triplet trials was significantly higher than across-triplet trials in Lag2 (t(14)=2.63, p< .05), but not in Lag8 (t(14)=0.86, p>.05). The interaction between lag and condition was marginally significant. No such pattern was observed for the Random group. The results suggest that implicitly learned temporal structure among targets attenuates the AB effect.
Meeting abstract presented at VSS 2016