Chapter 2 Selectivity in Social Information Processing
2.2 Study 1
The primary aim of Study 1 was to provide an initial test of whether status triggers differential motivations, which then heightens accessibility of motivationally relevant concepts. Additionally, the current study was also interested in whether there would a differentiation between social and non-social threat related words as a function of status.
Participants completed the status manipulation task, after which they completed a lexical decision task (LDT) that included social threat words, non-social threat words, and matched control neutral words. The LDT is a commonly used reaction-time measure of accessibility. In this task, participants are asked to decide whether or not letter strings form English words. Crucially, some of the letter strings are target-related words, some are target-unrelated and others are non-words. Higher accessibly of the target concept is indicated by the relatively faster accurate detection of target-related words (Meyer & Schvaneveldt, 1971). This procedure allowed the assessment of the extent to which participants were quicker to respond to social threat and non-social threat words in comparison to the matched control-neutral words. This pattern of effects will inform us as to whether or not accessibility of threat-related concepts vary as a function of status. In addition to measuring accessibility, I also used the LDT to measure social goal activation. Using this implicit measure allowed us to identify whether low social status increases accessibility of social threat without the participants’ conscious control. As active goals can temporarily increase the cognitive accessibility of goal-relevant information (Shah, Friedman, & Kruglanski, 2002), I expected low status participants to identify words associated with social-threat compared to non-social threat and neutral words more quickly.
Method
Participants and design. Sixty-three participants recruited from the University College London (UCL) subject pool received monetary compensation for participation.
One participant with an overall mean accuracy lower than 3 SD of the sample was excluded. The remaining 62 participants (40 females, 22 males, Mean age = 23.89) were randomly assigned to a high-status condition (n =22), low-status condition (n =21) or control condition (n =19). This study employed a 3 (Status: high, low, control) x 3 (Word type: social threat, non-social threat, neutral) mixed-model design, with repeated measures on the last factor. There were no effects of participant gender, which are not discussed further.
Materials. A total of 32 social threat, 32 non-social threat and 64 neutral words (See Appendix I) were selected from the ANEW (Bradley & Lang, 1999) database.
Social threat, non-social threat and neutral words were matched in terms of length and frequency (See ion of lengths were constructed.
Table 2.1). Using the ARC nonword database (English) (Rastle, Harrington, &
Coltheart, 2002), 128 nonwords that matched the target words on distribution of lengths were constructed.
Table 2.1
Mean (and standard deviations) of lexical characteristics of word stimuli used in Study 1
Word type
Neutral Social threat Non-social threat
Word length 6.84 (1.48) 6.25 (1.59) 6.72 (1.22) Word frequency 26.77 (27.17) 30.39 (15.49) 20.45 (12.93)
Procedure. After providing informed consent, participants completed the entire experiment in individual cubicles. The study was described as a study on “linguistic
processing” and social factors that may influence this process. There were two parts to this study. Firstly, participants completed the status manipulation task. Next, they completed the lexical decision task.
Status manipulation task. The status manipulation was adapted from Kraus, Côté, and Keltner (2010). In this task, participants were presented with a ladder with 10 rungs. Participants were asked to think of the ladder as representing where people stand in their country. Participants received the following instructions:
“Now, please compare yourself to the people at the very bottom (top) of the ladder. These are people who are the worst (best) off – those who are least (most) respected in your society. In particular we’d like you to think about how you are different from these people in terms of your own social prestige. Where would you place yourself on this ladder relative to these people at the very bottom (top)?”
To strengthen the status manipulation, participants were asked to write about a recent interaction with a person from the bottom (top) of the ladder. In particular, participants were asked to “think about how the differences between you might impact what you would have talked about, and how the interaction went. ” Next, participants indicated their perceived position on the ladder relative to the person at the very bottom (top) (1 = bottom rung to 10 = top rung) (see Appendix II).
In the control condition, participants were not presented with the ladder. To match the recall task of the experimental conditions they were asked to write about their day yesterday.
Lexical decision task. Participants were tested individually in a cubicle. Each trial began with a fixation cross that appeared at the middle of the screen for 500ms followed immediately by the letter string. The letter string remained on the screen for 2000ms or until a response was given (whichever was earlier). Participants were
instructed to press the z key on the keyboard (marked green) if the string was a word, and the m key on the keyboard (marked red) when the string was not a word.
Participants were told to respond as quickly and as accurately as possible. There was a blank inter-trial interval of 1000ms after participants made a response or time-out (See Figure 2.1). After twelve practice items, the 128 words and 128 non-words were presented in a random order. Upon completion, participants were checked for suspicion, carefully debriefed, paid and thanked.
Figure 2.1 Trial sequence of lexical decision task in Study 1. Trial onset was indicated by a fixation cross. This is followed by the presentation of a letter string (in this example, a social threat word). Participants then indicated using labelled keys on the keyboard whether the letter string was a word or a non-word.
Results and Discussion
Manipulation check. Participants’ indication of their perceived standing on the ladder served as the manipulation check scores; the bottom rung was coded as “1”, and the top rung was coded as “10”. The manipulation check scores were subjected to an
independent-samples t-test. Participants in the high-status condition reported significantly greater perceived status (M = 6.82, SD = 1.50) compared to participants in the low-status condition (M = 5.52, SD = 1.54), t (41) = -2.79, p < .01, d = .85. This indicates that the status manipulation was successful.
Reaction times. The data analysis for this task was based on RTs for correct responses. RTs faster than 200ms were excluded from analysis. Outliers, defined as RTs that deviated more than three SD from the overall mean RT were removed. Data from trials with errors and outliers were discarded and not analysed further. Error rates did not differ across status conditions, p = .702.
My primary interest was whether status influences the accessibility of neutral, social threat and non-social threat words. Shorter RT for social threat and/or non-social threat words as compared to matched control-neutral words would indicate that accessibility of those word categories had been activated as a result of the status manipulation task. For each participant, I averaged (separately) their RTs for each word category. Low-status participants were quickest to respond to social threat words (M = 549.84, SD = 66.57), followed by non-social threat words (M = 557.12, SD = 68.20), and neutral control words (M = 563.73, SD = 67.93). High-status participants were quickest to respond to non-social threat words (M = 547.43, SD = 54.61), followed by neutral control words (M = 556.74, SD = 61.12), and social threat words (M = 558.49, SD = 65.49). Control participants were quickest to respond to social threat words (M = 575.77, SD = 87.31), followed by neutral control words (M = 588.08, SD = 81.47), and non-social threat words (M = 590.26, SD = 75.77). These averages were subjected to a 3 (Status: high status, low status, control) x 3 (Word category: social threat, non-social threat, neutral) mixed-modal ANOVA with repeated measures on the latter factor.
Results revealed a main effect of word category, F (2, 118) = 3.32, p < .05, ηp2
= .053
such that participants were quickest to respond to social threat words (M = 560.86, SD
= 72.75), followed by non-social threat words (M = 563.84, SD = 67.64) and neutral words (M = 568.71, SD = 70.25). Of interest, there was a significant interaction between word category and status, F (4, 118) = 3.18, p < .02, ηp2
= .097. (See Figure 2.2).
Planned comparisons using one-way ANOVA for each status condition respectively revealed that there was a significant difference in accessibility across word categories for low status participants, F(2, 40) = 4.57, p < .02, ηp2
= .186. Low-status participants were marginally faster at categorising social threat words (M = 549.84, SD
= 66.57) compared to non-social threat words (M = 557.12, SD = 68.20), t (20) = -1.78, p = .09. Importantly, they were significantly faster at categorising social threat words compared to neutral words (M = 563.73, SD = 67.93), t (20) = -3.05, p < .01; the accessibility of non-social threat words did not differ significantly from neutral words, p > .2. There was no significant difference in accessibility across the word categories in the high-status and control condition.
Figure 2.2. Mean lexical decision RT for social-threat, neutral, and non-social threat words as a function of status conditions; error bars represent standard error.
The current findings provide preliminary support to the hypothesis that subjective perception of low-status heightens accessibility of social threat such that low-status participants were quicker at accurately identifying social threat words compared to matched neutral control words. High-status and control participants did not show this pattern of response as they did not differ in their response times across the word categories. Based on the overall pattern of response within each status category, it is worth noting that while low-status and control participants were quickest to respond to social threat words, high-status participants were slowest to respond to this category. This raises the possibility that high-status individuals may be inhibiting the accessibility of social threat words, however the current results lack statistical evidence for this claim.
In so far as cognitive accessibility of goal-relevant information informs the underlying social goal, the current study suggests that low-status individuals may be motivated by the self-protection goal, specifically in the social domain and not threat in general. The next study sought to investigate if this heightened accessibility for social threat extends to face perception. Specifically, if targets associated in a social threat context would be privileged in face processing. That is, the next study aimed to test the notion that heightened accessibility for social threat observed in low-status individuals would also lead to better memory of social stimuli (faces) associated with social threat.