Biasing spatial attention with semantic information: an event coding approach

Abstract

We investigated the influence of conceptual processing on visual attention from the standpoint of Theory of Event Coding (TEC). The theory makes two predictions: first, an important factor in determining the influence of event 1 on processing event 2 is whether features of event 1 are bound into a unified representation (i.e., selection or retrieval of event 1). Second, whether processing the two events facilitates or interferes with each other should depend on the extent to which their constituent features overlap. In two experiments, participants performed a visual-attention cueing task, in which the visual target (event 2) was preceded by a relevant or irrelevant explicit (e.g., “UP”) or implicit (e.g., “HAPPY”) spatial-conceptual cue (event 1). Consistent with TEC, we found relevant explicit cues (which featurally overlap to a greater extent with the target) and implicit cues (which featurally overlap to a lesser extent), respectively, facilitated and interfered with target processing at compatible locations. Irrelevant explicit and implicit cues, on the other hand, both facilitated target processing, presumably because they were less likely selected or retrieved as an integrated and unified event file. We argue that such effects, often described as “attentional cueing”, are better accounted for within the event coding framework.

Appendix

In this section, we report analyses of RT data from both experiments, without the exclusion of RT outliers. Here, we excluded the incorrect trials and trials immediately following an incorrect response. Response times, therefore, could take any value up to 2000 ms, which was the cut-off built-in of the procedure.

Experiment 1

Mean RTs from Experiment 1 were submitted to a 2 × 3 × 2 × 2 repeated measures ANOVA, with factors being cue Relevance, cue-target SOA, cue-target compatibility, and cue-target distance. This analysis revealed a main effect of SOA (F[2,38] = 71.84, p < 0.001, η ²_p  = 0.791). As SOA increased, RTs decreased (484, 377, and 340 ms, respectively, for SOAs = 300, 550, and 800 ms). We also found a main effect of Compatibility (F[1,19] = 6.10, p = 0.023, η ²_p  = 0.243), driven by faster responses with cue-target compatible than incompatible trials (respectively, 391 ± 27 and 409 ± 31 ms). Both of these results matched the findings reported in the main text after outlier-exclusion. The only other significant finding was a two-way interaction between Relevance and SOA (F[2,38] = 4.04, p = 0.026, η ²_p  = 0.176). The effect of SOA was more pronounced with irrelevant cues (514, 400, and 345 ms, respectively, for SOAs = 300, 550, and 800 ms), compared to relevant cues (453, 355, and 335 ms). No other main effect or interaction reached statistical significance. Most importantly, we found no interaction between Relevance x Compatibility (F[1,19] < 0.48). In short, the main results of Experiment 1 (i.e., the main effect of cue-target Compatibility) persist in the absence of the RT outliers-detection/exclusion procedure.

Experiment 2

Mean RTs from Experiment 2 were submitted to the same repeated measures ANOVA. This analysis also revealed a main effect of SOA (F[2,38] = 122.29, p < 0.001, η ²_p  = 0.866). As SOA increased, RTs decreased (491, 381, and 352 ms, respectively, for SOAs = 300, 550, and 800 ms). We also found a two-way interaction between relevance and distance (F[1,19] = 7.56, p = 0.013, η ²_p  = 0.285), driven by faster responses when the target appeared near the relevant (i.e., selected) word compared to when the target appeared near the irrelevant (i.e., not selected) word (401 vs. 419 ms). Finally, we found a three-way interaction between relevance, compatibility, and distance (F[1,19] = 4.65, p = 0.044, η ²_p  = 0.197). Matching the results reported in the main text, we found the same pattern of Relevance x Compatibility interaction—i.e., facilitation with cue-target compatibility when the cue is irrelevant, but an inverse compatibility when the cue is relevant—only with cue-target near trials (39 ± 22 ms, Cohen’s d = 0.39), but no such interaction when the target and cue appear far from each other (−5 ± 12 ms, Cohen’s d = 0.09). No other main effect or interaction reached significance. Thus, the central result of Experiment 2 (i.e., the three-way interaction between relevance, compatibility, and distance) remains the same without the application of the RT outliers-detection/exclusion procedure.