Physical aggression appears to have a much stronger effect on behavior than its overt frequency of occurrence would suggest. Studies examining effects of observing aggressive behavior and others looking at cognitive differences related to aggressive behavior suggest that physical aggression might be processed preferentially by the cognitive system. In order to examine this hypothesis, adult participants were given a social cue learning task. Study 1 presented three conditions which were identical except for the description of one potential outcome. Participants learned the cue more often when this outcome involved physical aggression (hit) compared to non-physical aggression (steal) or being friendly. Study 2 showed that the specific nature of the cue had no effect on success rates. Study 3 compared unusual outcomes involving aggression (being hit with a suitcase) and non-aggression (dancing). Results show that people were more accurate with the former. These results provide evidence that physical aggression is processed more efficiently than other forms of social interaction and gives a cognitive basis for the specific impact of physical aggression on behavior.
Understanding aggressive behavior and its impact is a continuing problem on many levels. Both the variety of theoretical accounts of aggression, varying from evolutionary (Archer, 1988 and Archer, 2009) to social theories (Anderson and Bushman, 2002 and Bandura et al., 1961) and the prevalence of aggressive behaviors bear witness to its importance. Psychological research into aggressive behavior has concentrated on two major themes, both of which imply that aggressive behavior has a particularly strong influence on the cognitive system. A first strand of research has shown that witnessing violent behavior in the community (Margolin & Gordis, 2004; Spano, Rivera, & Bolland, 2006) and in more indirect settings such as television and video games (Anderson, 2004; Anderson et al., 2003) is related to increases in aggressive behavior. The second strand of research has been developed in the context of Social Information Processing theory (SIP) (Crick and Dodge, 1994 and Dodge et al., 2003). This theory suggests that aggressive individuals construct internal models of aggressive behaviors that are inaccurate, leading them to interpret a larger class of behaviors as aggressive, thus creating vicious cycles of aggressive behaviors and reactions (Dodge et al., 1990 and Dodge et al., 2003). Both of these strands of research suggest that aggressive behavior has a particularly high level of salience, compared to other forms of social behavior. This is illustrated by research showing that incorporating a single aggressive child into a group of less aggressive children produces an overall increase in levels of aggression (Dishion, McCord, & Poulin, 1999).
In fact, a straightforward cost–benefit analysis suggests that physical aggression should be particularly salient. Physically aggressive behavior can have an immediate negative impact, with the possibility of varying degrees of personal harm. Other kinds of aggressive or prosocial behaviors have less immediate consequences and more variability in outcomes. Physically aggressive behavior is potentially dangerous in the short-term, and being able to predict the occurrence of such behavior would have a direct impact on survival. Thus, it should be particularly useful to process social cues predictive of physically aggressive behavior more efficiently than cues related to other forms of behavior.
We examine this hypothesis by using a social cue learning task. This involves sequences of animated interactions in which the participant takes the viewpoint of a person who has left their cell phone on a bench, and approaches another person who has taken the phone. In each sequence, the animated character with the phone looks and dresses differently, and uses a variety of gestures. At some point in each sequence, the character will either (A) put two hands down with fists clenched or (B) put both hands behind their back. We compare three conditions, all of which use exactly the same visible information. For half the sequences in each condition, the character gives the phone back to the participant. The cue for this outcome is the character putting both hands behind their back. In the Hit condition, the alternative outcome leads to the participant being hit by the character. In the Steal condition, the character runs off with the phone. In the Talk condition, the character gives the phone back and is friendly. In all three conditions, the cue for the alternative outcome is the character putting two hands down with fists clenched. Participants view the potential outcomes fully only once before the learning trials. During the learning phase, they are given only incomplete sequences, with no outcome. Participants are instructed to discover the cue that determines what the outcome of each sequence would be. After each trial (during which they observe only the initial part of the interaction), participants predict the outcome, following which the computer indicates the actual outcome (by a short verbal description). Participants are also given the chance to indicate whether or not they thought that they had discovered the cue.
Our hypothesis was that cue learning would be more efficient when the alternate outcome involved physical aggression (the Hit condition).
