The encoding and decoding of liking from behavioral cues in both auditory and visual channels

Abstract

Fifty high school students met in same-sex dyads for the first time over a cup of coffee in an experimental room designed as a comfortable living room. They then responded to a questionnaire designed to measure liking and perceived similarity. The interactions were video-taped. Two different panels of 6 judges later either viewed (video only) or heard (audio-only) the videotape and raed the subjects behavior on a number of scales. The judges also judged the degree of liking felt by the subjects by estimating subjects responses to the questionnaire. In addition gaze behaviour during the interaction was measured. Factor analysis demonstrated that 60% of the variance in “The Liking Scale” was attributed to 14 items relating to liking and perceived similarity. A separate factor to assess perceived similarity could not be found. Twenty-nine percent of the variance in the liking reported by subjects was predicted by an interrelated pattern of expressing behaviour including approproate looking, mutual gaze, self-disclosure, synchrony in movement and gesture, expressiveness of the face and liveliness of the voice. Video judges liking correlated 0.33 with subjects liking and 48% of the variance in their judgements was explained by the valid cues of looking and expressiveness of the face. Audio judges liking, although it correlated at 0.34 with video judges liking, did not correlate at all with subjects' liking because of an over-reliance on the important content cues. It is suggested that major problems leading to decoding inaccuracy may be over-reliance on content cues and over-confidence in the possibility of decoding accuracy. The evidence suggests that differences between subjects in encoding may be considerable.