Gilead et al. propose that abstract representation enables predictive cognition. Applying this idea to two major areas in social psychology – attitudes and close relationships – generates novel insights and hypotheses for the science of human evaluation and interpersonal liking. In particular, Gilead et al.'s framework points to an important distinction between experienced evaluations (e.g., “Right now, I like this tall man”) and abstract ideas about liking (e.g., “Generally, I like tall men”). Furthermore, their arguments suggest that abstract ideas about liking enable predictive cognition and mental travel.
Attitude researchers and close relationships researchers have not directly made the distinction between experienced evaluations versus abstract ideas about liking. In this commentary, we leverage Gilead et al.'s framework to highlight important new directions for each research area (Fig. 1).
Figure 1. Liking in two literatures. Ideas about liking (top) are more abstract than experiences of liking (bottom) because they treat individual exemplars as substitutable.
First, attitude researchers often study liking for social and nonsocial categories using measures that focus on liking for the overall category (e.g., a person's evaluation of the category “African Americans”) as well as measures that focus on liking for individual exemplars of that category (e.g., the average of a person's evaluations of a series of individual African American faces). In Gilead et al.'s language, evaluations of categories versus exemplars can be arranged along a continuum of mental abstraction: People's ideas about how much they like broad social categories are more abstract than their experienced evaluations of specific exemplars. Yet researchers have typically treated evaluations of exemplars and evaluations of categories as conceptually equivalent (see e.g., Ajzen & Fishbein Reference Ajzen and Fishbein1977, who treated both as measures of so-called “general attitudes;” see Cooley & Payne Reference Cooley and Payne2019, for a notable exception), which seems problematic in light of Gilead et al.'s framework. For example, Gawronski (Reference Gawronski2019) pointed out that the literature on implicit bias has largely ignored confounds between type of measure (implicit versus explicit) and target object (exemplars versus categories). Whereas explicit bias measures typically involve evaluating categories, implicit bias measures involve evaluative responses to exemplars.
Gilead et al.'s arguments further suggest that category evaluations, like other abstract representations, may function to enable predictive cognition. Whereas experienced evaluations of specific exemplars may guide immediate decisions about what to do in the here-and-now, abstract evaluations of overall social categories may guide decisions about what to do at a spatially distant location, in a hypothetical scenario, or in the future (e.g., who to hire to fill a future position). Moreover, abstract evaluations may be especially useful for making predictions about situations one has not yet experienced and that may be difficult to simulate (e.g., whether to move to a new city with a particular set of demographics).
Second, human mating researchers often study liking for attributes using measures that focus on liking for an overall attribute (e.g., a person's evaluation of the trait “intelligence” in a romantic partner) as well as measures that focus on liking for individual exemplars that vary in terms of a given attribute (e.g., the extent to which intelligence in a series of individual potential partners drives a person's evaluation of each partner). Again, these evaluations can be arranged along a continuum of mental abstraction. People's ideas about how much they like an attribute as a general concept are more abstract than their experienced evaluations of specific exemplars that embody those attributes. Yet researchers have typically treated them as conceptually equivalent (for reviews, see Eastwick et al. Reference Eastwick, Luchies, Finkel and Hunt2014; Ledgerwood et al. Reference Ledgerwood, Eastwick and Smith2018), which again seems problematic in light of Gilead et al.'s framework. For example, Ledgerwood et al. (Reference Ledgerwood, Eastwick and Smith2018) observed that the literature on human mating has largely ignored the distinction between evaluative experiences of traits (e.g., people's evaluations of romantic partners who are more vs. less intelligent) and evaluations of traits in the abstract (e.g., people's evaluations of the trait intelligence in a partner).
Distinguishing between abstract and concrete attribute evaluations has allowed us to ask new questions about how people form abstract representations of liking for attributes. For example, our research suggests that people form abstract attribute preferences by drawing on not only their concrete evaluative experiences, but also incidental features of the learning context (Eastwick et al. Reference Eastwick, Smith and Ledgerwood2019; Wang et al. Reference Wang, da Silva Frost, Eastwick and Ledgerwood2020). These incidental contextual features include how plentiful a trait is (e.g., whether the potential mates that one encounters are generally high or low in intelligence) and how much liking someone is generally experiencing (e.g., whether the potential mates that one encounters are generally desirable or undesirable). Thus, people's ideas about how much they like various traits in the abstract may be biased by the context in which they learn about their likes and dislikes.
Furthermore, abstract attribute preferences, like abstract category evaluations, may serve the critical purpose of enabling predictive cognition. For example, humans may rely on their abstract attribute preferences to predict whether they will like a potential date (or friend or colleague) who is spatially distant rather than close, or hypothetical rather than real (see Eastwick et al. Reference Eastwick, Finkel and Eagly2011; Huang et al. Reference Huang, Ledgerwood and Eastwick2020). Moreover, abstract attribute preferences may be especially useful for making predictions about situations that one learns about through socially acquired knowledge, rather than direct experience, and that may therefore be difficult to simulate (e.g., whether to visit a bar that a friend describes as full of particularly quirky patrons). Consistent with this notion, recent research suggests that abstract, summarized preferences primarily predict situation selection at a distance (e.g., whether to sign up for a dating website described as featuring highly intelligent partners) rather than situations that have been directly experienced (e.g., whether to sign up for a dating website after experiencing example profiles of potential partners that look highly intelligent; Wang et al. Reference Wang, da Silva Frost, Eastwick and Ledgerwood2020).
Going forward, we urge scholars to more seriously distinguish between abstract ideas about liking and concrete experiences of liking. If abstraction enables predictive cognition, as Gilead et al. posit, this distinction may prove both crucial and generative.
Gilead et al. propose that abstract representation enables predictive cognition. Applying this idea to two major areas in social psychology – attitudes and close relationships – generates novel insights and hypotheses for the science of human evaluation and interpersonal liking. In particular, Gilead et al.'s framework points to an important distinction between experienced evaluations (e.g., “Right now, I like this tall man”) and abstract ideas about liking (e.g., “Generally, I like tall men”). Furthermore, their arguments suggest that abstract ideas about liking enable predictive cognition and mental travel.
Attitude researchers and close relationships researchers have not directly made the distinction between experienced evaluations versus abstract ideas about liking. In this commentary, we leverage Gilead et al.'s framework to highlight important new directions for each research area (Fig. 1).
Figure 1. Liking in two literatures. Ideas about liking (top) are more abstract than experiences of liking (bottom) because they treat individual exemplars as substitutable.
First, attitude researchers often study liking for social and nonsocial categories using measures that focus on liking for the overall category (e.g., a person's evaluation of the category “African Americans”) as well as measures that focus on liking for individual exemplars of that category (e.g., the average of a person's evaluations of a series of individual African American faces). In Gilead et al.'s language, evaluations of categories versus exemplars can be arranged along a continuum of mental abstraction: People's ideas about how much they like broad social categories are more abstract than their experienced evaluations of specific exemplars. Yet researchers have typically treated evaluations of exemplars and evaluations of categories as conceptually equivalent (see e.g., Ajzen & Fishbein Reference Ajzen and Fishbein1977, who treated both as measures of so-called “general attitudes;” see Cooley & Payne Reference Cooley and Payne2019, for a notable exception), which seems problematic in light of Gilead et al.'s framework. For example, Gawronski (Reference Gawronski2019) pointed out that the literature on implicit bias has largely ignored confounds between type of measure (implicit versus explicit) and target object (exemplars versus categories). Whereas explicit bias measures typically involve evaluating categories, implicit bias measures involve evaluative responses to exemplars.
Gilead et al.'s arguments further suggest that category evaluations, like other abstract representations, may function to enable predictive cognition. Whereas experienced evaluations of specific exemplars may guide immediate decisions about what to do in the here-and-now, abstract evaluations of overall social categories may guide decisions about what to do at a spatially distant location, in a hypothetical scenario, or in the future (e.g., who to hire to fill a future position). Moreover, abstract evaluations may be especially useful for making predictions about situations one has not yet experienced and that may be difficult to simulate (e.g., whether to move to a new city with a particular set of demographics).
Second, human mating researchers often study liking for attributes using measures that focus on liking for an overall attribute (e.g., a person's evaluation of the trait “intelligence” in a romantic partner) as well as measures that focus on liking for individual exemplars that vary in terms of a given attribute (e.g., the extent to which intelligence in a series of individual potential partners drives a person's evaluation of each partner). Again, these evaluations can be arranged along a continuum of mental abstraction. People's ideas about how much they like an attribute as a general concept are more abstract than their experienced evaluations of specific exemplars that embody those attributes. Yet researchers have typically treated them as conceptually equivalent (for reviews, see Eastwick et al. Reference Eastwick, Luchies, Finkel and Hunt2014; Ledgerwood et al. Reference Ledgerwood, Eastwick and Smith2018), which again seems problematic in light of Gilead et al.'s framework. For example, Ledgerwood et al. (Reference Ledgerwood, Eastwick and Smith2018) observed that the literature on human mating has largely ignored the distinction between evaluative experiences of traits (e.g., people's evaluations of romantic partners who are more vs. less intelligent) and evaluations of traits in the abstract (e.g., people's evaluations of the trait intelligence in a partner).
Distinguishing between abstract and concrete attribute evaluations has allowed us to ask new questions about how people form abstract representations of liking for attributes. For example, our research suggests that people form abstract attribute preferences by drawing on not only their concrete evaluative experiences, but also incidental features of the learning context (Eastwick et al. Reference Eastwick, Smith and Ledgerwood2019; Wang et al. Reference Wang, da Silva Frost, Eastwick and Ledgerwood2020). These incidental contextual features include how plentiful a trait is (e.g., whether the potential mates that one encounters are generally high or low in intelligence) and how much liking someone is generally experiencing (e.g., whether the potential mates that one encounters are generally desirable or undesirable). Thus, people's ideas about how much they like various traits in the abstract may be biased by the context in which they learn about their likes and dislikes.
Furthermore, abstract attribute preferences, like abstract category evaluations, may serve the critical purpose of enabling predictive cognition. For example, humans may rely on their abstract attribute preferences to predict whether they will like a potential date (or friend or colleague) who is spatially distant rather than close, or hypothetical rather than real (see Eastwick et al. Reference Eastwick, Finkel and Eagly2011; Huang et al. Reference Huang, Ledgerwood and Eastwick2020). Moreover, abstract attribute preferences may be especially useful for making predictions about situations that one learns about through socially acquired knowledge, rather than direct experience, and that may therefore be difficult to simulate (e.g., whether to visit a bar that a friend describes as full of particularly quirky patrons). Consistent with this notion, recent research suggests that abstract, summarized preferences primarily predict situation selection at a distance (e.g., whether to sign up for a dating website described as featuring highly intelligent partners) rather than situations that have been directly experienced (e.g., whether to sign up for a dating website after experiencing example profiles of potential partners that look highly intelligent; Wang et al. Reference Wang, da Silva Frost, Eastwick and Ledgerwood2020).
Going forward, we urge scholars to more seriously distinguish between abstract ideas about liking and concrete experiences of liking. If abstraction enables predictive cognition, as Gilead et al. posit, this distinction may prove both crucial and generative.