Sticky plans: Inhibition and binding during serial-task control

Abstract

Recent evidence suggests substantial response-time costs associated with lag-2 repetitions of tasks within explicitly controlled task sequences [Koch, I., Philipp, A. M., Gade, M. (2006). Chunking in task sequences modulates task inhibition. Psychological Science, 17, 346-350; Schneider, D. W. (2007). Task-set inhibition in chunked task sequences. Psychonomic Bulletin & Review, 14, 970-976], a result that has been interpreted as inhibition of no-longer relevant tasks. Experiments 1-3 confirm much larger lag-2 costs under serial-control than under externally cued conditions, but also show (a) that these costs occur only when sequences contain at least two distinct chunks and (b) that direct lag-2 repetitions are not a necessary condition for their occurrence. This pattern suggests the hypothesis that rather than task-set inhibition, the large lag-2 costs observed in complex sequences, reflect interference resulting from links between positions within a sequential plan and the individual tasks controlled by this plan. The remaining experiments successfully test this hypothesis (Experiment 4), rule out chaining accounts as a potential alternative explanation (Experiment 5), and demonstrate that interference results from information stored in long-term memory rather than working memory (Experiment 6). Implications of these results for an integration of models of serial-order control and serial memory are discussed.

Figure 1

Sequence of events in the task-span procedure.

Figure 2

Experiment 1: Mean RTs and error percentages for each sequence position and the cued task selection condition and as a function of lag-2 repetitions versus changes.

Figure 3

Experiment 1: Mean RTs for the regular-sequence trials as a function of sequence grammar and position.

Figure 4

Experiment 2: Mean RTs and errors for the sequenced and cued trials of the hybrid sequences and as a function of lag-2 repetitions versus changes. The letters beneath the X-axis indicate if the current and preceding tasks were randomly cued (C) or sequenced (S). As a comparison also the data for Experiment 1 are shown, collapsed across the first and the second chunk.

Figure 5

Experiment 3: Mean RTs and error rates for the sequenced trials as a function of lag-2 repetitions versus changes. Note, that lag-2 repetitions for positions 1 and 2 are not actual repetitions, but relative to the previous sequenced positions (i.e., spanning 2–6 cued trials). Triangles for position 1 indicate mean RTs and errors for the ABA-CBC (inverted) and the ABC-ACB (upright) sequences (discussed in section “Self-Inhibition?”).

Figure 6

Experiment 4: Mean RTs and error rates as a function of chunking patterns, position, and lag-2 repetitions versus changes.

Figure 7

Experiment 4: Empirical RTs and model RTs as a function of chunking patterns, sequence grammars, and position. Filled arrows indicate lag-2 repetition trials and unfilled arrows indicate trials where positional interference should be large.

Figure 8

Experiment 5: Mean RTs and error rate as a function of sequence position and lag-2 repetitions versus changes. The unique task element was used for sequence positions 1 to 3. The dashed line shows RTs for trials with the unique task and which were always lag-2 changes.

Figure 9

Experiment 6: Mean RTs and error rate as a function of chunk repetition versus switch, sequence position, and lag-2 repetitions versus changes. Note, that lag-2 repetitions were defined relative to the alternative chunk, even for chunk-repeat trials.