A Sequential Learning Analysis of Decisions in Organizations to Escalate Investments Despite Continuing Costs or Losses

Abstract

Reinforcement process may underlie decisions frequently found in organizations to escalate investments of time, money and other resources in strategies (e.g., product development, capital investment, plant expansion) that do not result in immediate reinforces. Whereas cognitive biases have been proffered in previous explanations, the present analysis suggested that this persistence is a form of resistance to extinction arising from experiences with past investments that were variably reinforced. This explanation was examined in two experiments by varying the pattern of returns and losses subjects experienced for investment decisions prior to experiencing a series losses. Consistent with the proposed explanation, two conditions resulted in higher levels of recommitment during continuous losses: (a) training using a variable schedule of partial reinforcement, and (b) no training on the task. Results indicate that behavior analysis can be used to understand and control situations in organizations that are prone to escalation, such as investments in the research and development of new product lines and extensions of further loans to customers.

Intermittent Reinforcement and the Persistence of Behavior: Experimental Evidence

“…The fact that our results are consistent with experiments conducted on persistence and escalation under different conditions (see, e.g., Golz, 1992;Brecher & Hantula, 2005) 19 speaks, in our view, to the strength of the reinforcement manipulation in our experiment and rules out the need to postulate social explanations such as self-justification (Brockner, 1992). Note, however, that we are not arguing that processes such as self-justification do not occur when persistence is observed in naturally occurring environments.…”

“…For example, in a field study, Deslauriers and Everett (1977) showed how, to increase bus ridership, it was more effective to offer passengers variable as opposed to continuous reinforcement (gifting tokens to roughly every one in three passengers versus every passenger receiving a token). Golz (1992) pioneered an experimental game in which groups of participants could earn or lose money by investing under continuous and variable reinforcement conditions. In the acquisition phase, the market allowed participants to be successful; in the extinction phase, it did not.…”

The Costs of Eating: A Behavioral Economic Analysis of Food Refusal

“…In addition, this approach permits a priori prediction of responding, reinforcement, and punishment in situations of choice (Herrnstein, 1970(Herrnstein, , 1990. Finally, a behavioral economic framework allows integration of these clinical assessment and treatment data with the growing synthesis of behavioral and economic principles (Akerlof, 1991;Becker, 1976;Goltz, 1992;Hantula & Crowell, 1994;Hursh, 1980;Tustin, 1994) predicted by Kagel and Winkler (1972) over two decades ago.…”

Microworlds for experimental research: Having your (control and collection) cake, and realism too

“…The INVE$TMENT CHOICE$ MW (Hantula & Crowell, 1994a, 1994b) was used to investigate how changes in the patterning of investment return rates affected investment decisions during trials in which subjects made money or steadily lost money (this research replicated and extended work by Goltz, 1992, who used a similar MW task). The continued investment of resources despite steady losses has been dubbed "escalation and persistence of commitment" in the organizational psychology literature (Hantula, 1992;Staw & Ross, 1989).…”

“…Single decisions may be the product of an entirely different class of variables than are repeated decisions (Sitkin & Pablo, 1992). It is entirely probable that distinct levels of horizontal analysis may exist (such as a single decision, two decisions in sequence, three in sequence, and greater than three in sequence), and with them issues ofpattern and cyclicity (Goltz, 1992;Hantula & Crowell, 1994b), chaos (Barton, 1994), "strings" of gains and losses (Rachlin, 1990;Rachlin & Siegel, 1994), and temporal discounting (Ainslie, 1992;Kirby & Marakovic, 1995), which ought to be considered in a useful understanding of phenomena.…”