Attentional control transfers beyond the reference frame

Note that Wendt, Kluwe, and Vietze (2008) presented PC-unbiased items in novel locations in space; however, they were equidistant from both MC and MI items so those authors did not examine transfer of differential control settings to new locations.

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

As in Weidler and Bugg (2016), RTs from central locations were not of interest or analyzed in the primary analyses because RTs were expected to be relatively fast (and compatibility effects were expected to be reduced) given that the stimuli were presented at fixation (see Corballis, & Gratton 2003 for a similar pattern). For the interested reader, it is noted that RTs for unbiased trials in the central location were faster in compatible (M = 567 ms) than incompatible trials (M = 668 ms), t(59) = 20.41, p < .001, and there were fewer errors in compatible (M = 0.007) than incompatible (M = 0.028) trials, t(59) = 7.18, p < .001. As can be seen from the means, overall RTs were faster and flanker compatibility effects (101 ms) were reduced in the central location (because the stimuli appeared at fixation) compared to either the MC (184 ms) or MI (153 ms) training locations. These observations were confirmed by reliable main effects – F(1, 59) = 1166.04, p < .001, η_p² = 0.95 for MC and F(1, 59) = 1270.23, p < .001, η_p² = 0.96 for MI – and interactions, F(1, 59) = 240.32, p < .001, η_p² = 0.80 for MC and F(1, 59) = 147.98, p < .001, η_p² = 0.72 for MI – from 2 Location × 2 Compatibility ANOVAs that compared the central unbiased trials to the MC and MI trials, respectively, across all five test blocks. Neither overall RTs, F(1, 59) = 1.97, p = .166 for the main effect of phase, nor the compatibility effect differed as a function of experiment phase; F(1, 59) = 2.51, p = .119, for the interaction from a 2 Phase (training or transfer) × 2 Compatibility repeated measures ANOVA; M_training = 98 ms, M_transfer= 106 ms. Phase did produce a main effect in the same analysis on error rate, F(1, 59) = 6.70, p = .012, η_p² = 0.10 (M_training = 0.013, M_transfer= 0.021), and marginally interacted with compatibility, F(1, 59) = 3.06, p = .086, η_p² = 0.05, as compatibility effects were larger in the transfer phase (0.024) than the training phase (0.017).

As in Experiment 1, there was an effect of compatibility for the central location in both RT, t(59) = 23.53, p < .001(M_compatible = 532 ms, M_incompatible = 630 ms) and error rate, t(59) = 5.64, p < .001(M_compatible = 0.004, M_incompatible = 0.024). Also as in Experiment 1, RTs were faster in the central locations than in either the MC, F(1, 59) = 1209.53, p < .001, η_p² = 0.95, or MI location, F(1, 59) = 1289.66, p < .001, η_p² = 0.96, and the compatibility effect was reduced in the central location (98 ms) compared to either the MC (177 ms), F(1, 59) = 317.26, p < .001, η_p² = 0.84, or MI (147 ms), F(1, 59) = 99.81, p < .001, η_p² = 0.63, location. Furthermore, neither overall RTs nor the compatibility effect in the central location differed as a function of phase, Fs < 1 (M_training = 98, M_transfer = 100.) In the same analysis error rate phase produced a marginally reliable main effect, F(1, 59) = 3.03, p = .087, η_p² = 0.049, with more errors in the transfer phase (0.015) than training phase (0.012). Phase and compatibility did not interact in the error data in Experiment 2 (F < 1).

As in Experiments 1 and 2 there was an effect of compatibility for the central location in both RT, t(59) = 28.55, p < .001(M_compatible = 528 ms, M_incompatible =  627 ms) and error rate, t(59) = 7.33, p < .001(M_compatible = 0.003, M_incompatible = 0.035). Also as in Experiments 1 and 2, RTs were faster in the central locations than in either the MC, F(1, 59) = 1756.36, p < .001, η_p² = 0.97, or MI location, F(1, 59) = 1281.16, p < .001, η_p² = 0.96, and the compatibility effect was reduced in the central location (99 ms) compared to either the MC (170 ms), F(1, 59) = 355.41, p < .001, η_p² = 0.86, or MI (136 ms), F(1, 59) = 93.41, p < .001, η_p² = 0.61, location. In addition, an analysis examining the compatibility effect as a function of phase revealed a significant Phase × Compatibility interaction, F(1, 59) = 6.85, p = .011, such that the compatibility effect was larger during the training phase (105 ms) compared to the transfer phase (94 ms). This may reflect practice. The same analysis on error rate phase produced a significant main effect of phase, F(1, 59) = 5.99, p = .017, η_p² = 0.09, with more errors in the transfer phase (0.014) than training phase (0.010). Phase and compatibility did not interact in the error data in Experiment 3 (F < 1).

For thoroughness, in this first counterbalance there was also a reliable main effect of phase, F(2, 58) = 12.35, p < .001, η_p² = 0.30 (M_{Training phase} = 702 ms, M_{Transfer phase 1} = 679 ms, M_{Transfer phase 2} = 669 ms). This was qualified by a marginally significant Phase × PC interaction, F(2, 58) = 3.06, p = .055, η_p² = 0.10 (M_{Training phase MI−MC} = 5 ms, M_{Transfer phase 1 MI−MC} = -1 ms, M_{Transfer phase 2 MI−MC} = 11 ms), as well as a Phase × Compatibility interaction, F(2, 58) = 5.39, p = .007, η_p² = 0.16 (M_{Training phase I−C} = 162 ms, M_{Transfer phase 1 I−C} = 156 ms, M_{Transfer phase 2 I−C} = 138 ms).

Also, in the second counterbalance, there was a reliable main effect of phase, F(2, 58) = 8.20, p = .001, η_p² = 0.22 (M_{Training phase} = 673 ms, M_{Transfer phase 1} = 651 ms, M_{Transfer phase 2} = 645 ms). This was qualified by a marginally significant Phase × PC interaction, F(2, 58) = 2.71, p = .075, η_p² = 0.09 (M_{Training phase MI−MC} = -6 ms, M_{Transfer phase 1 MI−MC} = 6 ms, M_{Transfer phase 2 MI−MC} = 1 ms), as well as a Phase × Compatibility interaction, F(2, 58) = 8.78 p < .001, η_p² = 0.23 (M_{Training phase I−C} = 162 ms, M_{Transfer phase 1 I−C} = 152 ms, M_{Transfer phase 2 I−C} = 136 ms).

There was also a main effect of phase, F(2, 58) = 5.38 p = .007, η_p² = 0.16 (M_{Training phase} = 0.025, M_{Transfer phase 1} = 0.033, M_{Transfer phase 2} = 0.038). The phase × compatibility interaction was significant, F(2, 58) = 4.50 p = .015, η_p² = 0.13 (M_{Training phase I−C} = 0.046, M_{Transfer phase 1 I−C} = 0.060, M_{Transfer phase 2 I−C} = 0.067).

The analysis revealed no main effect of phase (F < 1), but a Phase × PC interaction, F(2, 58) = 4.84, p = .011, η_p² = 0.14 (M_{Training phase MI−MC} = − 0.024, M_{Transfer phase 1 MI−MC} = − 0.002, M_{Transfer phase 2 MI−MC} = − 0.008).