Fig 1.
Category structures for the six experiments in order from L to R.
Datasets 1 and 2 used the leftmost category structure (A). Datasets 3–6 used the remaining category structures (B-E). Features outlined in grey were irrelevant for categorization. Note that in Dataset 1, the two A categories have different relevant features than the two B categories. In Dataset 2, the irrelevant features were consistent for all categories. For Datasets 3 and 4, there was an additional manipulation of instructions to emphasize speed or accuracy. For Datasets 5 and 6, the rule-based categories are shown on top and the information-integration categories are shown on the bottom. Features were broadly similar to those used in Experiment 2, but were manipulated in Photoshop to have continuous values (size, curvature and rotation). There was an additional feature dimension not shown (thus, three in total) that was irrelevant for categorization.
Fig 2.
From left to right: fixation cross, response phase, feedback phase. The right panel shows the feedback button style used in all Datasets except Dataset 2 (5:1&1:1-4cat).
Fig 3.
Dataset 2 (5:1&1:1-4cat), a gamepad was used to record inputs, and no response boxes were shown.
For Feedback, which indicated the participants response and the correct response, was displayed in the centre of the screen as shown.
Fig 4.
Response time during stimulus presentation and during feedback by trial bin.
Each bin contains 24 trials and error shading represents SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.
Fig 5.
Mean time (per trial) spent fixating each area of interest (feedback button, irrelevant stimulus feature, relevant stimulus feature) for trials prior to reaching learning criterion.
Conditions that were not significantly different were collapsed. Light bars represent the second condition (i.e., Accuracy, II). Error bars represent SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.
Fig 6.
Probability of fixating each area of interest (feedback button, relevant stimulus feature, irrelevant stimulus feature) on each of the first three fixations of a trial.
This is conditioned over all trials, whether or not they had three fixations. Conditions that were not significantly different were collapsed. Dashed lines represent the second condition (i.e., Accuracy, II). Shaded area represents SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.
Fig 7.
Differences in the allocation of attention between the response and feedback phase plotted over trial bins by dataset and condition.
Shaded areas represent the SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.
Fig 8.
Differences in the proportion of attention allocated to the irrelevant feature between the response and feedback phase plotted over trial bins by dataset and condition.
Shaded areas represent the SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.
Fig 9.
Mean difference (error trials minus correct trials) in total time fixating each AOI (feedback button, relevant stimulus feature, irrelevant stimulus feature), plotted by dataset and condition.
Conditions that were not significantly different were collapsed. Lighter bars represent the second condition (i.e., Accuracy, II). Error bars represent SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.
Fig 10.
The probability of clicking through the feedback phase with no fixations, plotted by trial bins and dataset.
Shaded area represents SEM. In top-to-bottom order, the left-hand column contains Datasets 1 and 2, the middle column contains Datasets 3 and 4, and the right-hand column contains Datasets 5 and 6.