Getting it right generally, but not precisely: learning the relation between multiple inputs and outputs.

In real-world situations, people are often faced with the complex task of deciding which of many potential variables are affecting their own or others' behavior, as well as noting which specific aspects of behavior are being affected. Although it is common for professionals who encounter such conditions to claim that they acquire accurate and specific knowledge from their experience, it is unclear that such confidence is justified. Using a managerial task, we examined participants' ability to learn how various interventions affect various aspects of their employees' performance. The results of three experiments reveal that although participants appear to avoid prescribing an intervention that has a positive effect on a primary performance measure and a negative side effect on a secondary measure, when asked directly about the impact of the intervention, they respond by reducing their judgments of its positive impact. This was true regardless of whether participants indicated clear knowledge of its negative side effect (Experiment 3) or did not (Experiments 1 and 2). Thus, participants appear to be automatically integrating across the effects on different outcome measures.

A motivationally-based simulation of performance degradation under pressure.

The CLARION cognitive architecture has been shown to be capable of simulating and explaining a wide range of psychological tasks and data. Currently, two theories exist to explain the psychological phenomenon of performance degradation under pressure: the distraction theory and the explicit-monitoring theory. However, neither provides a detailed mechanistic explanation of the exact processes involved. We propose such a detailed theory within the CLARION cognitive architecture that takes into account motivation and the interaction between explicit and implicit processes. We then use our theory to provide a plausible explanation of some existing data. The data are simulated using the theory within the CLARION cognitive architecture.

Facilitative interactions of model- and experience-based processes: implications for type and flexibility of representation.

People are often taught using a combination of instruction and practice. In prior research, we have distinguished between model-based knowledge (i.e., acquired from explicit instruction) and experience-based knowledge (i.e., acquired from practice), and have argued that the issue of how these types of knowledge (and associated learning processes) interact has been largely neglected. Two experiments explore this issue using a dynamic control task. Results demonstrate the utility of providing model-based knowledge before practice with the task, but more importantly, suggest how this information improves learning. Results also show that learning in this manner can lead to "costs" such as slowed retrieval, and that this knowledge may not always transfer to new task situations as well as experientially acquired knowledge. Our findings also question the assumption that participants always acquire a highly specific "lookup" table representation while learning this task. We provide an alternate view and discuss the implications for theories of learning.

Developing rich and quickly accessed knowledge of an artificial grammar.

In contrast to prior research, our results demonstrate that it is possible to acquire rich, highly accurate, and quickly accessed knowledge of an artificial grammar. Across two experiments, we trained participants by using a string-edit task and highlighting relatively low-level (letters), medium-level (chunks), or high-level (structural; i.e., grammar diagram) information to increase the efficiency of grammar acquisition. In both experiments, participants who had structural information available during training generated more highly accurate strings during a cued generation test than did those in other conditions, with equivalent speed. Experiment 2 revealed that structural information enhanced acquisition only when relevant features were highlighted during the task using animation. We suggest that two critical components for producing enhanced performance from provided model-based knowledge involve (1) using the model to acquire experience-based knowledge, rather than using a representation of the model to generate responses, and (2) receiving that knowledge precisely when it is needed during training.

Effects of model-based and memory-based processing on speed and accuracy of grammar string generation.

Learners are able to use 2 different types of knowledge to perform a skill. One type is a conscious mental model, and the other is based on memories of instances. The authors conducted 3 experiments that manipulated training conditions designed to affect the availability of 1 or both types of knowledge about an artificial grammar. Participants were tested for both speed and accuracy of their ability to generate letter sequences. Results indicate that model-based training leads to slow accurate responding. Memory-based training leads to fast, less accurate responding and highest achievement when perfect accuracy was not required. Evidence supports participants' preference for using the memory-based mode when exposed to both types of training. Finally, the accuracy contributed by model-based training declined over a retention interval.