Cars as Cues to Person Perceptions – Who Said What? Paradigm

Background Human faces possess cues that signal social information about a wide variety of identities, but non-person objects such as automobiles can also convey important social information. In a previous study conducted by our lab, we found that participants made gendered inferences about drivers from their vehicles. Specifically, participants perceived trucks and dark-colored vehicles as being driven by more masculine than feminine drivers, and small and light-colored vehicles as being driven by more feminine than masculine drivers. These findings suggest that vehicles are encoded with social category information. The present study seeks to provide process-level evidence that vehicles activate gender categories that are retained in memory. This study will employ a modified version of the Who Said What? paradigm (WSW: Taylor et al., 1978). Whereas the WSW paradigm typically pairs statements with pictures of people and participants’ task is to remember “who said what?”, we will instead present pictures of vehicles along with statements describing the behavior of each vehicle’s driver. In an initial learning task, participants’ will view a series of vehicles, one at a time, along with a statement describing the driver’s behavior. The vehicles were rated by participants in a previous study in this line of research as signaling that their drivers are either highly masculine or highly feminine (i.e., Cohen’s d ≥ .65). We pretested the behavioral statements to ensure that participants perceive the described behaviors as neither masculine nor feminine, for example “drove past a bank”. Next, participants will complete a distractor task consisting of the Personal Attributes Questionnaire (Spence et al., 1973) and the Need for Cognition scale (Lord & Putrevu, 2005) After completing the distractor task, participants will view a series of behavioral statements, one at a time, along with an array of the eight vehicles that were presented in the learning task. Their task here is to select which (if any) of the vehicles was originally paired with the behavioral statement. In addition to the previously presented statements, participants will also view new statements that were not presented with any vehicle during the learning task. The correct response to these new statements is to select “None. This statement is new” rather than assigning it to any of the vehicles. Predictions & Analysis Plan Our hypothesis is that vehicles activate social categories, and consequently social category information will influence participants’ responses in the absence of memory for specific behavior-vehicle pairings. As initial evidence of social categories influence on participants’ memory, we predict that participants will make more within-category errors than between-category errors. For example, if a participant does not correctly remember which specific feminine-signaling vehicle a behavioral statement was paired with, they should be more likely to mistakenly assign it to another feminine-signaling vehicle (i.e., a within-category error) than to a masculine-signaling vehicle (i.e., a between-category error). To test this prediction, we will compute within-category and between-category errors for each vehicle type for each participant. Because the correct response is by definition a within-category vehicle, within-category errors have a lower base rate probability (3) than do between-category errors (4). Consequently, we will apply Pietraszewski’s (2018) correction by multiplying between-category errors by (n-1)/n, with n denoting the number of exemplars per category, which translates to ¾ in this case. Then, we will compare within- versus (corrected) between-category errors using a within-participants t-test. Additionally, to provide relatively more direct evidence of social category influence on participants’ memory, we will apply Klauer and Wegener’s (1998) multinomial processing tree model for the WSW paradigm. This model separately estimates the contributions of item memory (i.e., for specific behavioral statements), speaker memory (i.e., the specific vehicle with which the behavioral statement with paired), category memory (i.e., the social category associated with the specific vehicle), and various guessing processes. We predict that category memory will significantly contribute to responses, such that the confidence intervals around parameters reflecting category memory will not include zero. We also have a series of exploratory predictions. Our first exploratory prediction is that the magnitude of effects will be moderated by the driver gender signaled by the vehicle. In previous studies in this line of research, participants’ ratings of the extent to which vehicles signaled the masculinity of their drivers were descriptively higher than were participants' ratings of the extent to which vehicles signaled the femininity of their drivers. Similarly, in pretesting behavior statements, participants’ ratings of behaviors that signaled male drivers were descriptively higher than were participants’ ratings of behaviors that signaled female drivers. Based on these observations, we are tentatively developing a hypothesis that vehicles that signal masculine drivers do so more strongly than do vehicles that signal feminine drivers. However, we have competing predictions about the direction of effects implied by this asymmetry. One possibility is that the stronger gender signaled by masculine-associated vehicles provides more individuating information, which would manifest in terms of fewer within- versus between- category errors, higher speaker memory (“speaker” refers to “vehicle” in the context of our modified WSW task), and/or lower category memory in comparison to feminine-associated vehicles. Alternatively, another possibility is that the stronger gender signaled by masculine-associated vehicles will override individuation and induce reliance on categorization, which would manifest in terms of more within- versus between-category errors, lower speaker memory, and/or higher category memory in comparison to feminine-associated vehicles We will test these competing predictions with exploratory analyses that separate responses as a function of masculine- versus feminine-associted vehicles. Additionally, we will include the Personal Attributes Questionnaire (Spence et al., 1973) and the Need for Cognition scale (Lord & Putrevu, 2005) as filler tasks for participants to complete before the first and second parts of the WSW task. The Personal Attributes questionnaire assesses gender stereotype endorsement, and the Need for Cognition scale assesses dispositional tendencies to engage in careful thinking. We will explore whether either of these scales moderate our predicted WSW effects, with the idea that people who more strongly endorse gender roles and/or are less dispositionally inclined to think carefully will rely on gender categories more and therefore will show larger WSW effects. Participant Sampling Information Based on the power analysis reported by Heidrich and Imhoff (2025), and accounting for some attrition and inattentive responding, we will recruit 300 U.S.-based participants from CloudResearch Connect. Exclusion criteria Our experimental paradigm includes a number of questions designed to screen out bots and identify participants who are responding inattentively. Additionally, the platform where we will host the experiment (i.e., Qualtrics) offers a bot-detection tool. This tool quantifies the likelihood that a respondent is a human versus bot on a [0,1] scale with scores greater than .5 meaning that the respondent is likely a human. We will remove a priori the data of all participants who fail any of our attention checks, as well as any respondents who score