Executive Summary
Expertise derived from experience is widely regarded as valuable. Managers, market mavens, and applicants to MBA programs rightly view their years of experience as an important credential. Moreover, experience is viewed as intrinsically different from the type of “book learning” that occurs in school. However, experiential learning in the marketplace does not occur for its own sake or for the sake of achieving high evaluations from an instructor. Rather, learning frequently occurs as a by-product of repeatedly making decisions about concrete actions and then observing the outcomes. Thus, such learning is action based.
In three experiments, the authors contrast two broad classes of decision tasks that arise when managers or consumers engage in action-based learning about environmental relationships between multiattribute stimuli and a continuous criterion. Numerical prediction tasks require decision makers to use observed attributes to predict the exact value of a continuous criterion. For example, an investor interested in real estate might review the specifications of a particular property and attempt to estimate both its current market value and its value at some time in the future. Categorical prediction tasks require decision makers to decide whether the observed attributes indicate that the continuous criterion is above or below a given cutoff. For example, a person shopping for a new home might review the same information as the real estate investor to decide whether a particular property warrants further investigation (i.e., calling his or her agent, arranging a tour). Later, after screening many properties, this same person might engage in a prediction task to make an offer on a specific property.
As the example demonstrates, an important issue is how learning transfers across tasks. Learning and transfer across related tasks is likely to play a role when consumers construct consideration sets and later evaluate the considered options or when consumers engage in price search at one price level but then are “upsold” by a salesperson to a higher price level. This type of transfer is also common in managerial decision making; managers are frequently called on to take actions and make predictions within a given environment (e.g., the low end of the product line) and then, months or years later, must shift to a different environment (e.g., the high end of the product line).
The authors demonstrate that action-based learning can be either accurate and efficient or errorful and biased. Action-based learning is also risky because decision makers face substantial risks of missing significant environmental relationships, overgeneralizing knowledge, and exhibiting poor transfer of learning, even when task differences are small. The authors propose a new model-based explanation for why these effects are observed. The attention-band model describes how goals affect learning through attention, and it explains why changes in attentional allocation provide efficiency gains in some situations and accuracy costs in others. Thus, the model allows the delineation of the circumstances under which decision makers will be most at risk.
The authors apply the attention-band model to the stimuli and use the results to make strong a priori predictions about how study participants will allocate attention and how performance will transfer from categorical to numerical prediction tasks, and vice versa. The results across all three experiments support the predictions. Both latency measures and diagnostic learning elements that were part of the design helped show that participants allocated increased attention, as predicted by the attention-band model. Furthermore, this allocation of attention materially affected learning, accuracy, and transfer across tasks. The results are incompatible with several other accounts of learning and attention.
Biography
Eric M. Eisenstein is Assistant Professor of Business in the S.C. Johnson School of Management at Cornell University. His areas of research focus on managerial and consumer decision making, expertise, learning, and intuitive statistics. He holds a PhD in Managerial Science and Applied Economics and an MA in Statistics from the Wharton School, University of Pennsylvania. Previously, he worked as a strategy consultant at Mercer Management Consulting. Dr. Eisenstein graduated from the Management and Technology dual-degree program at the University of Pennsylvania, where he concurrently earned a BS in Economics from the Wharton School and a BS in Computer-Systems Engineering from the School of Engineering and Applied Science
J. Wesley Hutchinson is Stephen J. Heyman Professor and Professor of Marketing in the Wharton School at the University of Pennsylvania. He received a PhD in Cognitive Psychology from Stanford University and a BS in Psychology from Duke University. His research interests focus on consumer and managerial decision making, particularly the interrelationships among attention, learning, confidence, decision making, and expertise in repeated choice environments. He has published articles in a wide variety of journals in business and psychology, including Journal of Consumer Research; Journal of Marketing Research; Marketing Science; Marketing Letters; Harvard Business Review; Psychological Review; Psychometrika; Journal of Experimental Psychology: Learning, Memory, and Cognition; and Memory & Cognition. He is a two-time winner of the Journal of Consumer Research Outstanding Article Award (1985–1987, 2000) and was a finalist for the Journal of Marketing Research O’Dell Award (1992–1994). He is on the editorial review boards of Journal of Consumer Research and Marketing Science (associate editor 1994–1999). He was president of the Association for Consumer Research (2003).
J Marketing Research, Volume 43, Number 2, May 2006
View Table of Contents.