Richard A. Carlson

Effects of repetition and foreknowledge in task-set reconfiguration.

To examine the roles of executive control and automatic activation in task switching, we manipulated foreknowledge as well as task transitions. In Experiments 1 and 2, performance with foreknowledge was faster than performance with no foreknowledge, but the amount of switch cost did not depend on foreknowledge. This result suggests that switch costs primarily reflect persisting activation rather than inadequate preparation. In Experiment 3, switch cost was greater with foreknowledge about task transition alone than with foreknowledge about both task transition and identity, suggesting that foreknowledge about specific task identity did allow preparation for a switched task. We argue that task repetition and foreknowledge effects are independent. Although foreknowledge allows preparation for both repeated and switched tasks, repeating the same task has benefits over task switching regardless of foreknowledge.

Practice and Working Memory Effects in Building Procedural Skill

Several theories assume that practice (a) results in restructuring of component processes and (b) reduces demand on working memory. Eight subjects practiced judgments about digital logic gates for over 8,000 trials. At two practice levels, subjects made judgments while retaining short-term memory loads irrelevant to the judgments, relevant but not accessed, or accessed to make the judgments. Four phenomena together provide constraints for theory: First, performance declined in moving from blocked practice to randomized practice. Second, gate and judgment type strongly affected latency. Third, these effects declined but did not disappear with practice. Fourth, the cost of accessing information in working memory remained substantial. These results are interpreted as reflecting a serial pro-cess with constant structure, while component processes become faster. The results challenge theories assuming that all learning results from restructuring or that restructuring is an automatic consequence of practice, and they support a distributed view of working memory. Several current theories of skill acquisition (e.g., Anderson, 1982, 1983, 1987; Rosenbloom & Newell, 1987) share the assumptions that (a) practice results in the reorganization or r£rtrwc/un/7# of component processes, and (b) practice reduces the load on a working memory that serves as a single workspace for carrying out cognitive processes. In the study reported here, we observed the acquisition of procedural skill for judgments about causal rules over the course of approximately 20 hr (over 8,000 trials) of practice. The results challenge the above assumptions and suggest an alternative view in which the organization of component processes and use of working memory remain constant while the speed of component processes increases (and attentional load decreases) with extended practice (see Schneider, 1985).

Practice Schedules and the Use of Component Skills in Problem Solving

In motor and verbal learning, random practice schedules produce poorer acquisition performance but superior retention relative to blocked practice. We extend this contextual interference effect to the case of learning cognitive procedural skills to be used in problem solving. Subjects in three experiments practiced calculation with Boolean functions. After this acquisition phase, subjects solved problems requiring these procedures. Experiments 1 and 2 demonstrated superior transfer to problem solving for skills acquired under random schedules. In Experiment 3, subjects practiced component skills in a blocked schedule, with one of four tasks—same-different judgment, mental arithmetic, short-term memory, or long-term memory—intervening between trials. For same-different judgments and mental arithmetic, transfer performance was comparable to that found for random schedules in Experiments 1 and 2. This result suggests that the differences depend on processing rather than storage demands of intertrial activity. Implications for theories of problem solving and part-whole transfer are discussed.

What do the hands externalize in simple arithmetic

In 4 experiments, the authors examined the use of the hands in simple arithmetic tasks. Experiments 1 and 2 demonstrated that pointing increases both accuracy and speed in counting arrays of items, whether those items are identical or distinctive. Experiment 3 demonstrated that individuals tend to nod their heads when not allowed to point and that nodding is associated with greater accuracy, suggesting that pointing is functional for reasons other than simply providing additional visual information. Experiment 4 examined changes in speech when adding arrays of digits, depending on whether participants were allowed to use their hands to manipulate the tokens on which the digits were presented. Taken together, the results of these experiments are consistent with recent research suggesting that gesture can serve cognitive functions and that the hands can support the binding of representational elements to their functional roles by providing phase markers for cyclic cognitive processes.

Strategy guidance and memory aiding learning a problem-solving skill

Guidance can help learners overcome the difficulties of getting started in a novel domain, but it is often ineffective in promoting learning and transfer. This article examines two aspects of guidance-communicating solution strategies for a problem domain and providing working memory support-in learning a novel problem-solving skill. Subjects in two experiments learned to troubleshoot simulated information networks. The learning environment varied in type of guidance provided-none, variable template, fixed template, and procedural instruction-and in availability of memory aiding. Variable-template guidance was effective when memory aiding was provided, and procedural instructions produced effective learning with or without memory aiding. However, fixed-template guidance was not effective, and there was no consistent effect of memory aiding in unguided, discovery learning conditions. The results have theoretical implications for the locus of guided-learning effects and suggest practical guidelines for the design of guided-learning environments.

Consistency and restructuring in learning cognitive procedural sequences.

Ss practiced mental arithmetic tasks that were consistent or varied at two levels: the sequence of operations and the data for those operations. Experiment 1 demonstrated separable benefits of sequence- and data-level consistency. In Experiment 2, sequence and data consistency varied within subjects, and the sequence of operations did not match the goal structure. Final performance benefited from consistent data but not from consistent sequence. In Experiment 3, sequence consistency varied between subjects. Sequence and data consistency had separable benefits

Coordinating information from perception and working memory.

Individuals must often coordinate information in working memory with information from perception. The demands of coordination have been analyzed in terms of the cost to switch attention. Coordination is considered in terms of the organization of control processes. Ss in 4 experiments performed list-processing tasks that sometimes required alternation between sets of items that were held in working memory or were currently displayed. Experiment 1 demonstrated that performance was slower and more error-prone when alternating between sets than when reporting a single set. On alternation tasks, latency varied with serial position, indicating that Ss paused between pairs of responses. In Experiment 2, this serial position function was observed for tasks requiring alternation between sets of information in the same modality (memory or perception). Experiments 3 and 4 demonstrate that this effect depends on the requirement to generate a new sequence of responses. A model of control processes for coordination is developed and tested.

Component practice and exposure to a problem-solving context

The component fluency hypothesis suggests that practicing component skills can increase the fluency of target task performance. Three experiments examined boundary conditions for this hypothesis. College students practiced individual Boolean functions, then solved transfer problems using these functions. In Experiment 1, varying the amount of component practice had no detectable effect on problem solving fluency. In Experiment 2, subjects had 480 trials of component practice, with exposure to the target task after 0, 240, or 480 trials. A fourth group had 192 trials of component practice. The results partially replicated Experiment 1: exposure to the target context after 240 trials produced the best final performance. Experiment 3 replicated Experiment 2, using a more sensitive target task and an additional group exposed to the target task after 120 trials of component practice. Results confirmed the importance of target task exposure. These findings suggest limits on the component fluency hypothesis.

Component fluency in a problem-solving context

Theories of cognitive skill suggest two hypotheses about component task practice. First, component practice increases the speed of executing component skills during problem solving. Second, component practice produces component skills that behave as encapsulated subroutines. Eight college students practiced making judgments about digital logic gates for 1360 trials. At two points during practice, they solved circuit problems that did or did not require logic gate knowledge. Time per move declined with problem-solving practice, but the effects of component training were ambiguous. However, teaching circuit functions did increase problem-solving speed. In a second experiment conducted at the end of practice, the same subjects solved problems designed to test whether the logic gate judgments acted as encapsulated subroutines. Component judgments were slower in the problem-solving context than in isolation, disconfirming this hypothesis. Taken together, the results indicate the need for a more detailed theoretical rationale for component drill.

Procedural frameworks for simple arithmetic skills

Rule-application skills such as simple arithmetic are often used as components of complex, goal-directed routines, and evidence suggests that goals to perform such skills can be instantiated in advance of information about specific operands. The procedural framework hypothesis is that goal instantiation evokes frameworks that guide the application of procedural knowledge, suggesting distinct processing roles for operator and operand symbols. In contrast, the uniform role hypothesis suggests that both types of symbols serve only as retrieval cues. Participants in 4 experiments solved simple Boolean or standard arithmetic problems. Serial display of problem elements showed a consistent solution-time benefit for operator-first displays compared with operands-first displays, supporting the procedural framework hypothesis for both new and highly practiced skills.