Figure 1
A flowchart of the preregistered experiments conducted as part of the current study following the initial experiment (Experiment 1.0). Experiments were carried out sequentially, top to bottom, depending on the decision taken after the preceding experiment – Experiments and results are shown in square boxes, and decision points are shown in diamond boxes. AV and V stand for audiovisual and visual, respectively.
Figure 2
Schematic representation of a trial for each Binding condition in Experiment 1.0, closely following the paradigm of Cinar et al., 2025 in their no-prioritization condition. The same shape-color combinations are used here as an example, but in the actual task, pairings were chosen at random. The textbox that participants saw during the probe is omitted from the figure.
Figure 3
Results of Experiment 1.0 showing the main effects of Serial Position (left) and Binding Condition (center) as well as their interaction (right). Note, however, that the interaction contributed only weakly to the model fit and was thus not further investigated. Mean accuracy (displayed as percent correct) per level of Serial Position and Binding are shown numerically above each point on the graph. Error bars show the 95% credible interval. Serial position increased in the expected pattern, substantiating Confirmatory hypothesis 1. Meanwhile the binding conditions were sufficiently different, producing a main effect of Binding. However, the underlying patterns were unexpected: UO-AV was actually larger than UO-V, and UO-V and SS-V were not meaningfully different from each other.
Figure 4
Schematic representation of a trial for each Condition in Experiment 1.3, using the same color-shape combinations as in Figure 2 for comparison purposes. The textbox that participants saw during the probe is omitted from the figure.
Figure 5
Results of Experiment 1.3 showing the main effects of Serial Position (left) and Binding Condition (center), as well as their interaction (right). Mean accuracy (displayed as percent correct) per level of Serial Position and Binding are shown numerically above each point on the graph. Error bars show the 95% credible interval. Serial position once again increased in the expected pattern, while the levels of Condition once again differed meaningfully, but contrary to expectation. This time, recall accuracy in the UO-AV-S condition was the lowest, drawing out the main effect, and demonstrating what appeared to be a recall cost of bimodal storage. Also, there was no bimodal presentation advantage, as UO-AV-P and UO-V did not differ. Inspecting condition differences at each serial position, however, revealed moderate evidence for a bimodal advantage at serial position 1 in the UO-AV-P condition. In contrast, bimodal costs were evident at serial position 1 to 3 in the UO-AV-S condition.
Figure 6
Schematic representation of a trial for each Condition in Experiment 1.4, using the same color-shape combinations as in Figures 2 and 4 for comparison purposes. The textbox that participants saw during the probe is omitted from the figure.
Figure 7
Results of Experiment 1.4 showing the main effects of Serial Position (left), Binding Condition (center), and their respective interaction (right). Mean accuracy (displayed as percent correct) per level of Serial Position and Binding are shown numerically above each point on the graph. Error bars show the 95% credible interval. Serial position again increased according to expectations. The levels of Condition once again differed meaningfully, such that recall accuracy for UO-V was highest, while recall accuracy for the audiovisual conditions (i.e., UO-AV-P and UO-AV-S), which did not differ from each other, was lower. Thus, there was no bimodal advantage, but rather a bimodal cost, if anything. Comparisons at each serial position mostly corroborated the pattern of bimodal costs, except at the last serial position, at which moderate evidence for a bimodal advantage (UO-AV-S > UO-V) was observed.
Figure 8
Schematic representation of a trial for each Condition in Experiment 1.5, using the same color-shape combinations as in the previous figures for comparison purposes. The textbox that participants saw during the probe is omitted from the figure.
Figure 9
Results of Experiment 1.5 showing the main effects of Serial Position (left) and Binding Condition (middle) as well as their interaction (right). Mean accuracy (displayed as percent correct) per level of Serial Position and Binding are shown numerically above each point on the graph. Error bars show the 95% credible interval. Serial position once again increased according to expectations. Recall accuracy for UO-AV-VisP was higher than UO-V, showing evidence for a bimodal advantage when the recall cue was visual. However, there was no difference between the UO-AV-VerbP and UO-V conditions, showing no bimodal advantage for verbal recall cues. Follow-up tests, resolving the interaction, revealed strong evidence for a bimodal advantage at serial position 1 and 4 with visual probe presentation (UO-AV-VisP > UO-V). With verbal cue presentation, recall accuracy was diminished or comparable for audiovisual compared to unisensory trials (UO-AV-VerbP < UO-V) at serial positions 1 to 3, showing evidence against a bimodal advantage. At serial position 4, however, recall performance in the UO-AV-VerbP exceeded that in unisensory trials, showing very strong evidence for a bimodal advantage with verbal cues.