Two studies of relatively aggressive and relatively unaggressive emotionally disturbed boys in residential treatment examined whether the more aggressive children exhibit either an attributional bias to infer hostility regardless of the nature of the social stimuli that they appraised or an actual ability to detect true instances of hostility. In both studies, the social stimuli consisted of photographs that each depicted one of four classes of affectively charged interpersonal situations (created by crossing the bipolar dimensions of positivity-negativity and dominance-submission). Study 1, which included 32 boys, varied both the nature of the social stimulus materials and the possible appraisals (response alternatives) available to the subject in a way that distinguished the effects of attributional bias from those of actual ability to identify a particular class of social stimuli. Study 2, which included 40 boys, relaxed the reponse constraints imposed on subjects in the first study in order to assess the appraisals of the four classes of social stimuli that relatively aggressive and relatively unaggressive boys formulated spontaneously. Findings from both studies indicated that an attributional bias to infer hostility from various classes of social stimuli became more marked as aggressiveness increased. Implications are discussed.Cognitively oriented personality theorists tions to aversive stimuli. According to (e.g.,

Reciprocity of attraction has been called a fundamental principle of social relations. However, correlational data provide only meager support for actual reciprocity. A formal analysis reveals that the reciprocity correlation contains a mixture of two different correlations: reciprocity at the individual level and reciprocity at the dyadic level. These two correlations can be estimated from a round robin design. Data from 48 college students and 336 dyads, measured at five time points, showed a -.210 individual level correlation and a .617 dyadic level correlation. The discussion of these results considers three issues: the necessity of an appropriate design, the specification of a theoretical and statistical model that includes two levels of analysis, and the use of an indirect estimation procedure to calculate the model parameters."What makes the lamb love Mary so?" The eager children cry. "Why, Mary loves the lamb, you know, And that's the reason why." (nursery rhyme) Reciprocity has long enjoyed considerable prestige and status in social psychology as a fundamental principle of interpersonal attraction. Its intuitive appeal aside, the prestige and status of this so-called fundamental principle is due largely to the fact that several theoretical viewpoints readily predict the occurrence of reciprocity. For instance, social exchange theories (e.g., Blau, 1964;Emerson, 1967; Homans, 1961) contend that the reinforcement contingencies of voluntary social interactions and relations will produce reciprocity. More specifically, social exchange accounts of reciprocity assume that we tend to like those who reward us and also that those who reward us tend to like us. Granting the validity of these two assumptions, the conclusion that we tend to like those who like us follows logically.Cognitive-consistency theories (e.g., Heider,

Two experiments investigated the selective influences of experimentally induced mood states on children's encoding and retrieval of affectively valent information. Experiment 1 revealed that a happy, compared to a neutral, mood during encoding facilitated recall of positive information; conversely, a sad encoding mood disrupted recall of positive material. A happy mood during retrieval also facilitated recall of positive information, but no other selective effects of retrieval mood occurred. Experiment 2 indicated that the negative mood of anger, like that of sadness, disrupted the encoding of positive information; unlike sadness, however, anger facilitated the encoding of negative material. Again, no selective effects of retrieval mood occurred. Overall the findings indicate that selective encoding and retrieval may contribute to children's cognitive ability to regulate mood states as well as other aspects of social learning and development.

Two studies assessed recognition memory of interpersonal traits that subjects had rated according to either private self-reference (Study 1) or public self-reference (Study 2). Both studies also administered the Self-Consciousness Scale, which permitted a dual classification of subjects according to private self-consciousness (high and low) and public self-consciousness (high and low). Study 1 revealed a private false alarms effect (FAE), the strength of which was moderated by private selfconsciousness, whereas Study 2 revealed a public FAE, the strength of which was moderated by public self-consciousness. From the convergent and discriminant evidence, two hypotheses received support-namely, that (a) individuals articulate both private and public components of the selfschema, and (b) private self-consciousness predicts the extent to which individuals articulate the private component, whereas public self-consciousness predicts the extent to which individuals articulate the public component.

The study tested recognition memory of trait adjectives that subjects had rated previously according to self-descriptiveness. Prior to the test of recognition memory, the Self-Consciousness Scale was administered to classify subjects as being either high or low in private self-consciousness (the disposition to introspect). Only among subjects high in private self-consciousness did the commission of "false alarms" (responding old to distractors, or new adjectives that the target list did not actually contain) increase from the least to most self-descriptive traits. Moreover, subjects high in private self-consciousness committed more false alarms to the most selfdescriptive traits than subjects low in private self-consciousness but fewer to nonself-descriptive traits. Results were interpreted as supporting the hypothesis that persons high in private self-consciousness have articulated the self-schema more extensively than persons low in private self-consciousness.Previous research has consistently revealed positive relations between private self-consciousness, that is, the disposition to introspect (e.g., Buss, 1980;Fenigstein, Scheier, & Buss, 1975), and both the accuracy and efficiency of processing self-referent information. For example, persons who often introspect (high in private self-consciousness) write longer selfdescriptions (Turner, 1978b), furnish more valid self-reports (Scheier, Buss, & Buss, 1978;Turner, 1978a), decide more quickly whether undesirable traits describe the self (Turner, 1978c), and recall more self-referent information under conditions of incidental learning (Hull & Levy, 1979, Experiment 1; Turner, 1980) than persons who seldom introspect (low in private self-consciousness).The articulation of the self-schema, which conceivably differs between persons who do and persons who do not introspect often, may explain the relations between private self-consciousness and both the accuracy and efficiency of processing self-referent information. A selfschema, structurally, organizes the attributes and features that define the self-concept and, functionally, controls the processing (e.g., encoding, storage, and retrieval) of self-referent information (e.g.