14 Miller

In both Pavlovian conditioning and human causal judgment, competition between cues is well known to occur when multiple cues are presented in compound and followed by an outcome. More questionable is the occurrence of competition between outcomes when a single cue is followed by multiple outcomes presented in compound. In the experiment reported here, we demonstrated blocking (a type of stimulus competition) between outcomes. When the cue predicted one outcome, its ability to predict a second outcome that was presented in compound with the first outcome was reduced. The procedure minimized the likelihood that the observed competition between outcomes arose from selective attention. The competition between outcomes that we observed is problematic for contemporary theories of learning. When a cue (i.e., an antecedent event) is followed by an outcome (i.e., a subsequent event), the association that is ordinarily formed may be assessed predictively (i.e., by presenting the cue and assessing whether participants predict the outcome) or diagnostically (i.e., by presenting the outcome and assessing whether participants diagnose the cue). Moreover, when two cues are presented together prior to an outcome, the cue with the higher predictive value ordinarily competes with the other cue for predicting the outcome. Examples of cue competition include overshadowing (Pavlov, 1927), blocking (Kamin, 1968), and the relative stimulus-validity effect (Wagner, Logan, Haberlandt, & Price, 1968; Wasserman, 1974). Cue competition occurs in both humans and animals (e.g., Balaz, Gutsin, Cacheiro, & Miller, 1982; Kamin, 1968; Matute, Arcediano, & Miller, 1996; Shanks, 1985; Wasserman, 1990) and is now an established phenomenon that is addressed by many models developed in areas as diverse as neural networks, animal conditioning, causal attribution, and categorization (e.g., Gluck & Bower, 1988; Kruschke, 1992; Rescorla & Wagner, 1972; Van Hamme & Wasserman, 1994). In contrast, much less attention has been given to competition between outcomes. Competition between outcomes is important because most neural networks and contemporary associative theories of learning anticipate an absence of such competition. That is, presentation of a cue should evoke the expectation of all outcomes that were previously paired with that cue. Despite this clear prediction, competition between outcomes has rarely been investigated. Some studies have demonstrated competition between multiple effects of a common cause in causal judgment tasks (Chapman, 1991; Matute et al., 1996; Price & Yates, 1993, 1995; Shanks, 1991; Waldmann & Holyoak, 1992, Experiment 2), but most of those studies presented effects before causes (i.e., diagnostic training). Consequently, the apparent competition between effects observed in those studies could be interpreted as competition between cues rather than between outcomes (e.g., Shanks & López, 1996). Moreover, in the few instances in which effects were not presented before causes (e.g., Matute et al.), effects and causes were presented simultaneously. Thus, even though all of those studies seemingly demonstrated competition between effects, none can be unambiguously interpreted as demonstrating competition between outcomes (i.e., events presented after their cues). In a previous study offering some evidence of competition between outcomes (Esmoris-Arranz, Miller, & Matute, 1997), we used rats as subjects and auditory stimuli as events. Cue-Outcome 1 pairings were followed by a phase in which the cue signaled an Outcome 1–Outcome 2 compound. At test, Outcome 2 was presented to determine if the rats behaved as if the cue had occurred. For this reason, before testing with Outcome 2, we paired the cue with footshock so that occurrence of the cue produced fear. Rats with initial cue → Outcome 1 experience showed less fear than rats lacking the initial cue → Outcome 1 pairings. Thus, this study seemingly demonstrated blocking between outcomes. However, this study has an alternative interpretation. According to G. Hall (personal communication, September 22, 1994), when the cue → footshock pairings were given, the cue might have activated representations of Outcome 1 and Outcome 2, which consequently could have competed at that time for association with the footshock. In this view, Outcome 1 and Outcome 2 could be interpreted as cues that competed for an association with the footshock outcome, rather than as outcomes that competed for diagnosing the cue. A second issue in interpreting this previous study arises from our testing by presenting Outcome 2. This constituted diagnostic testing (despite the training procedure being predictive), and thus did not directly test the critical prediction of associative theories that competition between outcomes would not occur with predictive testing (i.e., presentation of the cue). Moreover, in other work, we (Matute et al., 1996) observed competition between effects in humans when we tested diagnostically, but not when we tested predictively. This pattern of results suggests that there may be differences in stimulus competition between diagnostic and predictive testing; however, we did not address the focal issue here concerning competition between outcomes because we trained with simultaneous presentation of causes and effects. The present study assessed competition between outcomes using both predictive training and predictive testing procedures, so that the competing elements were unambiguously outcomes (i.e., subsequent events) during the treatment and testing phases. We used a blocking design in which a cue (C) predicted Outcome 1 (O1) in Phase 1 and a simultaneous compound of O1 and Outcome 2 (O2) in Phase 2. Moreover, the preparation avoided cover stories (such as were used in most prior research on this issue), thereby minimizing the role of extraexperimental knowledge. Toward this end, rats served as subjects. Another concern was to minimize the biological significance of the competing outcomes. Rescorla (1980b) suggested that biologically significant stimuli demand more attention than stimuli of low biological significance. Thus, if O1 were biologically significant (e.g., food or footshock), it might compete with O2 by distracting the subject from O2. Although perceptual distraction is a potential basis of outcome COMPETITION BETWEEN OUTCOMES Ralph R. Miller 1 and Helena Matute 2 1 State University of New York at Binghamton and 2 Universidad de Deusto, Bilbao, Spain Address correspondence to Ralph R. Miller, Department of Psychology, SUNY-Binghamton, Binghamton, NY 13902-6000; e-mail: rmiller@ binghamton.edu. Research Report PSYCHOLOGICAL SCIENCE Ralph R. Miller and Helena Matute VOL. 9, NO. 2, MARCH 1998 147 competition, we wanted to determine whether competition between outcomes could occur on the basis of associative competition rather than distraction. To minimize the possibility of competition between outcomes being due to distraction, we used outcomes that during training were of low biological significance. We achieved this by embedding the stimulus competition treatment in a sensory preconditioning procedure so that O2 was not made biologically significant until after the competitive training phase.