Facial and Autonomic Manifestations of the Dimensional Structure of Emotion

Ward M. Winton, Lois E. Putnam, and Robert M. Krauss

Columbia University

As every student learns in introductory psychology, the initial coherent theoretical response to the first question was formulated by William James (1884, 1890/1950). Arguing against the common-sense view that the visceral changes associated with particular emotions are the product of an internal, experiential state, James’s explanation, which has come to be known as the James-Lange theory, reversed the casual sequence. "Bodily changes follow directly the perception of the exciting fact and … our feeling of the same changes as they occur IS the emotion" (James, 1884, pp. 189-190).[p.196]

. . .

[p.200] The present study was designed to examine facial and physiological responding associated with both the intensive and evaluative dimensions of affective experience. Subjects viewed a series of 25 emotionally evocative slides in a paradigm similar to that of Buck and colleagues (Buck et al., 1972, 1974). Heart rate and skin conductance responses were recorded, as were self-reports of slide pleasantness. Facial expressions were covertly videotaped and were later shown to judges who rated both the pleasantness and intensity of the subjects' affective response.

As detailed below, the major deviation from the procedure used by Buck and colleagues concerned the quantification of the physiological responses. Most importantly, second-by-second changes in cardiac response to stimulus slides were examined, along with phasic electrodermal responses, in order to determine if cardiac activity, like judges' pleasantness ratings, would reflect the evaluative dimension, and if electrodermal responses would reflect the intensive dimension. Such a demonstration of dissociation of physiological response systems, or directional fractionation (e.g., Lacey & Lacey, 1958), would add to our understanding of the relationship between subjective emotional experience and its bodily expression, providing data relevant both to theories of emotion based on a construct of unidimensional, undifferentiated arousal and to contemporary versions of the James-Lange position.

Method

Twenty-four male Columbia undergraduates viewed the slides and were paid $4 for their participation. Data were not analyzed for four subjects. Two guessed that a hidden camera was present and two objected to the videotaping procedure after it was revealed to them.

Ninety undergraduates judged the subjects’ facial expressiveness by viewing the videotapes. These included 30 men and 30 women who judged edited segments that were each 30 sec. long, and 30 additional men who judged edited segments that were each 10 sec in length. Eighteen judges received course credit for their participation, and seventy-two were paid $3 each.

An additional 20 men and 20 women had previously received $2.50 each for participating in an initial slide selection procedure.

Ten color pictures in each of five categories were culled from various books and magazines. These categories were the same as those used by Buck and colleagues (1972, 1974) : Sexual photographs of women taken from Playboy magazine; Scenic depictions of landscapes, trees, and beaches; Pleasant pictures of adults and children at leisure; Unpleasant photographs from a medical text, showing people badly burned or mutilated; and Unusual pictures including cubist paintings and time-lapse photos. Slides made from these pictures were shown to judges who were asked to indicate which of the five categories was most appropriate for each slide. A total of 40 judges participated, in groups of 2 to 5 at a time. The five slides eliciting the highest interjudge agreement in each category were chosen as stimuli for the slide-viewing experiment. For each of the 25 slides selected, a minimum of 73 percent of the judges agreed on its categorization.

The subject was seated in a comfortable armchair in a moderately lit, air-conditioned, 3 x 2.5 m room. He sat approximately 1.5m from a 38x38cm rear-projection screen on which the slides were projected from an adjacent control room. Several elaborate-looking pieces of equipment rested on a table next to the screen. Among these were a white noise generator and a sham "distortion monitor" (Olson, 1978), which concealed a Panasonic WV-240p video camera. The video cable was entwined with several purposeless wires that led into the next room. The presence of this apparatus was explained to the subject by telling him that "in some sessions, we play white noise through this speaker, and then we have to use all this equipment here. But in today’s session, you’ll be listening to white noise through earphones, so we won’t be using any of this."

The subject was told that the experiment was on "physiological and vocal responses" to pictures of various kinds. He was instructed to describe verbally how each slide made him feel, first by rating its pleasantness on a 7-point Likert-type scale (depicted near the screen) and then by talking freely. He was told to refrain from speaking immediately after slide onset, and to wait for a signal light next to the screen before talking. This procedure was similar to that of Buck et al. (1972, 1974) and it ensured that the physiological records during the first 20 sec after slide onset would be uncontaminated by the effects of vocalization. In the Buck studies, however, subjects rated the slides after speaking freely about their feelings, not before. After explaining the slide-viewing procedure, the experimenter asked the subject to sign a form indicating that he understood his remarks would be recorded on audio tape. All subjects agreed to sign.

The experimenter then asked the subject to put on a pair of Sennheiser HD414 headphones. These headphones delivered 82 dB of continuous white noise, filtered to restrict its range to 150-600 Hz, in order to mask sounds accompanying slide projector operation. The experimenter the affixed electrodes (Beckman regular Ag-AgCl) to measure skin conductance and heart rate. For skin conductance, electrodes were attached to the distal phalanges of the index and middle fingers of the left hand, with K-Y Jelly serving as the electrolyte. For heart rate, electrodes were attached in a standard lead II configuration.

Subjects were then instructed to relax and sit quietly for 10 min. After this adaptation period, the experimenter reentered the room and told the subject he would now view five "practice" slides to clear up any questions he might have regarding the procedure. These slides were drawn from the original group of 50 slides described above. Practice slides had not elicited high interrater agreement, but were most often assigned to the Scenic, Pleasant, or Unusual categories. After administering the practice slides, the experimenter returned to the subject room to ensure that the subject fully understood his task. The experimenter then initiated the sequence of 25 stimulus slides, remaining in the control room for the duration of the slide-viewing period.

Each stimulus slide was projected for 30 sec, and the rating-response signal in the subject’s room was illuminated during the last 20 sec. Between stimulus slides, a blank slide was projected for 8 sec. Order of presentation of the 25 slides was determined by a Latin square design, such that each slide category occurred once per five-trial block. Two 5 x 5 Latin squares were used: half of the subjects viewed the order of slide presentation determined by one square, and half by the other.

After the final slide, the experimenter returned to the subject’s room, removed the electrodes and earphones, and debriefed the subject. The covert videotaping procedure was explained, and the subject was asked to sign a consent form permitting the videotapes to be used in the subsequent parts of the experiment. Two subjects declined to sign, and their tapes were erased.

Skin conductance was recorded on a Beckman 411 polygraph using a constant-voltage coupler, and EKG was recorded on a second polygraph channel as well as on a Sony TC-630 audio tape recorder for off-line analyses. The tape-recorded EKG and slide-onset signals were later played into a Hewlett-Packard 521 AR electronic counter and 561-B digital recorder which timed and printed R-R interval durations to the nearest msec.

An 8-cam timer and two model 131C Hunter timers controlled timing of stimulus onsets as well as event signals on the polygraph and audio tape recorder. White noise background was produced by a General Radio 1390-B Random Noise Generator, filtered by an Allison Laboratories Model 2BR Band Pass Filter, and calibrated using the C-scale of a General Radio 155C sound level meter equipped with a flat plate adaptor. Audio and video recordings of the subject were made on a Sony V0-2800 videocassette recorder.

The videotapes of the subjects were viewed by 30 male and 30 female judges, who rated the pleasantness and intensity of the subjects’ emotional reactions on 7-point Likert-type scales. Specifically, judges were instructed to "indicate on the appropriate scale how pleasant or unpleasant you think the viewer’s emotional reaction was, and how intense it was." These ratings will be subsequently referred to as facial pleasantness and facial intensity. Each judge viewed all 25 trials, in consecutive order, for each of two subjects. Each subject was viewed by a total of six judges, and raters participated in groups of one to five. These judgements were made by viewing the subjects’ facial reactions during the full 30 sec of slide presentation.

An additional 30 judges made their ratings by watching edited videotapes which showed only the facial expressions that occurred during the first 10 sec of slide presentation. Since no sex differences in decoding ability were found among the first group of judges, only males were involved in this 10-sec judgement task. In these 10-sec ratings, each judge viewed two subjects, and each subject was viewed by three judges.

The videotapes were edited so that judges could hear the sound of the slide projector changing, and they were instructed to watch the subject’s face carefully to observe his first reactions. Judges could not hear subjects describing their reactions, but they were told that this was the subjects’ task. Judges also were apprised of the five different slide categories.

It should be noted that the above procedures differed from those of Buck and colleagues (1972, 1974, 1977), whose analyses of expressiveness involved one judge per subject and one subject per judge. By increasing the number of judges per subject and subjects per judge, the present study attempted to unconfound sender and receiver characteristics.

In evaluating subjects’ electrodermal response to the stimulus slides, SCR magnitude in the first response period, or MAGI, was of primary interest. In addition, three other measures of electrodermal activity were analyzed: MAG2, magnitude of the largest SCR in the second response period (5-11 sec after slide onset); MAG, magnitude of the largest SCR in the entire 1-11 sec period; and SCL, the level of skin conductance measured at slide onset. Skin conductance records for one subject were not scorable due to equipment malfunction.

Prestimulus HR and average HR per second for 13 post-slide sec were computed from the printed R-R interval data. The triphasic cardiac response during the first 10 sec after slide onset consisted of a 1-sec deceleration (D1), an acceleratory limb (A1) peaking at approximately 4 sec, and an anticipatory deceleration (D2) at approximately 9 sec. This waveform is typical of cardiac responses during such intervals preceding expected stimuli and/or responses. All three peaks--D1, A1, and D2,-- were significant by t tests on HR changes from prestimulus during each post-stimulus second. A subsequent acceleration, still rising at 13 sec, was also significant by this test and was most likely produced by vocalizations following the rating signal (cf. Campos & Johnson, 1966; Libby et al., 1973, p. 288). Subsequent analyses treated four indices of cardiac activity on each trial: The D1, A1, and D2 peaks (i.e., HR change from prestimulus at post-stimulus sec 1, 4, and 9, respectively) and prestimulus HR.

For each facial response to a slide presentation, four measures of facial expressiveness were computed. These consisted of the judges’ mean ratings, based on the10- and the 30- sec videotaped segments, of facial pleasantness (PL10 and PL30, respectively), and of facial intensity (INT30). [p. 203]