Deliberation erodes cooperative behavior–even towards competitive out-groups, even when using a control condition, and even when eliminating selection bias

Jim A.C. Everett, Zach Ingbretsen, Fiery Cushman, Mina Cikara
Journal of Experimental Social Psychology, 73, 76-81

By many accounts cooperation appears to be a default strategy in social interaction. There are, however, several documented instances in which reflexive responding favors aggressive behaviors: for example, interactions with out-group members. We conduct a rigorous test of potential boundary conditions of intuitive prosociality by looking at whether intuition favors cooperation even towards competitive out-group members, and even in losses frames. Moreover, we address three major methodological limitations of previous research in this area: a lack of an unconstrained control condition; non-compliance with time manipulations leading to high rates of exclusions and thus a selection bias; and non-comprehension of the structure of the game. Even after eliminating participant selection bias and non-comprehension, we find that deliberation decreases cooperation: even in competitive contexts towards out-groups and even in a losses frame, though the differences in cooperation between groups was consistent across conditions. People may be intuitive cooperators, but they are not intuitively impartial.

Overview of Paper


A recent body of research suggests that, far from requiring effortful control, behaving prosocially arises from “processes that are intuitive, reflexive, and even automatic” (Zaki & Mitchell, 2013, p. 466). Several studies find that people tend to make prosocial decisions in economic games more quickly than selfish ones, and time- pressure increases the incidence of prosocial behavior (for an overview and meta-analysis see Rand, 2016). To explain this, the Social Heuristics Hypothesis (SHH: Rand, 2016; Rand et al., 2014) posits that the social strategies typically successful in daily life (e.g., cooperation) become automatized as default responses, and that deliberation can override these defaults to modify behavior.While meta-analytic work has provided strong support for the claim that manipulating reliance on intuition through time pressure encourages prosociality (Rand, 2016; Rand et al., 2014), a recent Registered Replication Report (RRR) by Bouwmeester and colleagues (2017) finds only mixed support. In this paper we consider two challenges to the idea that intuition favors cooperation—one methodological and one theoretical—and then provide new experimental evidence. We investigated whether intuition would favour cooperation and deliberation favour selfishness, even when interacting with outgroup members and even in loss-framed contexts. Moreover, we tested whether this would hold even when addressing methodological concerns levelled at work on intuitive cooperation. We did this by drastically reducing exclusion rates; ensuring that the exclusion procedure was equated across experimental conditions; and introducing a control condition with no instructions to either deliberate or act intuitively.


We had 1316 American participants complete the experiment online using Amazon Mechanical Turk, and used a fully between-subjects design where participants completed a Prisoner’s Dilemma and we manipulated (1) whether participants were encouraged to make their decision intuitively, through deliberation, or in the absence of either instruction; (2) the group membership of the other player; and (3) whether the PD was framed in terms of gains or losses. This study had a fully between-subjects design with 12 conditions: 2 (group membership of other player: in-group vs. out-group) x 2 (frame: gains vs. losses) x 3 (instructions: time delay vs. time pressure vs. unconstrained control). Our design, hypotheses, and analysis plan were all pre-registered at the Open Science Framework.


  1. Participants were more likely to cooperate under time pressure than under time delay, and marginally more likely to cooperate when given no instructions than under time delay. Although participants cooperated slightly more under time pressure than when given no instructions, this difference was not significant.
  2. There was a significant effect of decision frame whereby participants were more likely to cooperate under a gains frame than a losses frame
  3. There was a significant effect of group membership such that participants were more likely to cooperate with an in-group than an out-group member.
  4. We found that time pressure increased cooperation even for outgroups with a difference of 18 percentage points; just when analyzing losses frames with a difference of 19 percentage points; and especially for outgroup members under a losses frame, with a difference of 27 percentage points.


In this paper we investigated intuitive cooperation, considering both methodological and theoretical challenges to this work. Our findings make three main contributions to the literature:

  1. First, our work addresses concerns about selection bias – the asymmetric exclusion of high numbers of participants across conditions - that have plagued work on intuitive cooperation (Bouwmeester et al., 2017; Tinghög et al., 2013). Specifically, we drastically reduced participant exclusions from around 50% to just 2%, equated exclusion rates across the two active manipulation conditions, and ensured that all participants understood the structure of the game. Even when addressing these methodological issues we observed greater cooperation in the time pressure condition relative to the time delay condition, supporting the claims of the SSH. At the same time, however, by including a neutral control condition, we found that intuitive cooperation effects seem driven more by time-delay induced deliberation reducing cooperation than by time-pressure induced intuition increasing cooperation.
  2. Second, our work sheds light on intuitive cooperation in intergroup contexts. Contrary to traditional assumptions about implicit intergroup cognition, according to which people harbour intuitive biases that are particularly revealed in the absence of deliberative control, increased time pressure did not detectably increase or decrease intergroup bias. intuition increases prosociality but also fails to eradicate bias: Under time pressure participants cooperate with everybody more, but regardless of time pressure they cooperate with ingroups most of all. While we may be intuitive cooperators, we are not intuitively impartial.
  3. Third, by investigating the boundary condition of losses framing on the effects of time pressure, ours is among the first studies to look at intuitive cooperation in both gains and losses frames. We extended previous work by showing that time pressure encourages cooperation even in losses frames, and even towards outgroup members in losses frames, and that there was greater overall cooperation in the gains context relative to the losses. This is consistent with work supporting a self- oriented perspective on loss aversion, whereby an aversion to losses makes people demand more in interpersonal interactions (De Dreu, Emans, Vliert, & Carnevale, 1994; De Dreu, Emans, & Van de Vliert, 1992), increases cheating (Schindler & Pfattheicher, 2017), and reduces prosocial behaviour (Brewer & Kramer, 1986).