Fast deliberation is related to unconditional behaviour in iterated Prisoners’ Dilemma experiments

Eladio Montero-Porras, Tom Lenaerts, Riccardo Gallotti, Jelena GrujićScientific 2022 · DOI

TL;DR

By applying cognitive modeling to game theory experiments, researchers found that "unconditional" players—those who almost always cooperate or always defect—make decisions significantly faster than those who seek equality. The study reveals that fast deliberation isn't a sign of being "good" or "bad," but rather a sign of acting categorically rather than reacting to the specific behavior of others.

What Problem Does This Paper Address?

In social dilemmas like the Prisoner’s Dilemma, why do some people cooperate while others defect? While we know factors like gender, age, and social norms play a role, the underlying cognitive process remains a "black box."

Previous research attempted to use simple Reaction Time (RT) to explain social preferences, but the results were often contradictory: some studies claimed cooperators are faster (intuitive cooperation), while others claimed defectors are faster. This paper addresses this confusion by moving beyond raw RT. Instead, the authors use the Drift Diffusion Model (DDM) to decompose decision-making into specific cognitive components: initial bias, the speed of evidence accumulation (deliberation), and perceived difficulty.

Methods

The researchers re-analyzed data from three previous Iterated Prisoner’s Dilemma (IPD) experiments using two primary tools:

  1. Drift Diffusion Model (DDM): This is a mathematical framework that assumes decisions are reached by accumulating "evidence" over time until a threshold is hit. It provides three key parameters:
  2. $a$ (Threshold): How much evidence is needed (a proxy for caution or perceived difficulty).
  3. $v$ (Drift Speed): How fast evidence is gathered (the rate of deliberation).
  4. $z$ (Bias): The starting point of the decision process (initial inclination).
  5. Relative Allocation angle ($RA^\circ$): The authors introduced this as a proxy for Social Value Orientation. It measures an individual’s desired payoff allocation between themselves and others. An angle of $90^\circ$ represents pure cooperation, $0^\circ$ represents pure defection, and $45^\circ$ represents a preference for equality/reciprocity.

Datasets

The study utilized three distinct experimental datasets involving the Iterated Prisoner’s Dilemma:

  • Pairwise IPD (PIPD): Subjects played against a single partner (either fixed or shuffled).
  • Moore Network IPD (mNIPD): Subjects played in a network where each person had eight neighbors.
  • Von Neumann Network IPD (vnNIPD): Subjects played in a network with four neighbors.

These datasets allowed the authors to compare how decision-making changes as the social environment becomes more complex (from 1-on-1 to 1-on-8).

Key Insights and Ideas

  • The "Middle" is Slow: The most striking finding is that participants near the $45^\circ$ mark (those seeking equality or acting conditionally) had a drift speed ($v$) near zero. They are the "slow deliberators" because they are constantly weighing their actions against their opponents' previous moves.
  • Unconditional Behavior is "Fast": Participants at the extremes—the "always cooperators" and "always defectors"—accumulated evidence much faster. If your strategy is fixed regardless of what others do, your brain reaches a decision threshold much more quickly.
  • Selfishness is Cognitively Demanding: In several treatments, individualistic/selfish players had a higher decision threshold ($a$). This suggests they perceived the game as more "difficult" or required more internal evidence to justify defecting compared to cooperative players.
  • Context Dictates Bias: In fixed-partner settings, players had a higher initial bias ($z$) toward cooperation. In shuffled settings, where partners change constantly, that intuitive "pro-social" starting point vanished.

Key Results

  • Drift Speed as a Predictor: Drift speed ($v$) was the strongest variable for explaining the variance in social preferences ($RA^\circ$). In the pairwise shuffled experiment, the correlation was remarkably high (Spearman’s $\rho = 0.84$).
  • Fixed vs. Shuffled: Evidence collection is significantly faster when playing against a fixed neighbor than a shuffled one. Stability in social circles reduces the cognitive load of deciding whether to trust someone.
  • Network Effects: In network-based games (mNIPD), a significant portion of participants (16-31%) defected unconditionally, resulting in very low $RA^\circ$ scores and very fast decision times.

Limitations and Future Work

  • Binary Constraints: The experiments were limited to binary choices (Cooperate or Defect). Real-world social dilemmas often involve a spectrum of choices, which the current DDM setup doesn't fully capture.
  • Proxy vs. Direct Measure: While $RA^\circ$ is a clever proxy for social preferences, it is derived from behavior within the game rather than an independent psychological assessment.
  • Future Directions: The authors suggest applying this DDM approach to more complex games and exploring how humans interact with artificial agents (AI) that can react much faster than human counterparts. They also note that this modeling approach could replace expensive fMRI or eye-tracking setups for researchers wanting to "peek" into the deliberation process.

Summary updated

Abstract

People have different preferences for what they allocate for themselves and what they allocate to others in social dilemmas. These differences result from contextual reasons, intrinsic values, and social expectations. What is still an area of debate is whether these differences can be estimated from differences in each individual's deliberation process. In this work, we analyse the participants' reaction times in three different experiments of the Iterated Prisoner's Dilemma with the Drift Diffusion Model, which links response times to the perceived difficulty of the decision task, the rate of accumulation of information (deliberation), and the intuitive attitudes towards the choices. The correlation between these results and the attitude of the participants towards the allocation of resources is then determined. We observe that individuals who allocated resources equally are correlated with more deliberation than highly cooperative or highly defective participants, who accumulate evidence more quickly to reach a decision. Also, the evidence collection is faster in fixed neighbour settings than in shuffled ones. Consequently, fast decisions do not distinguish cooperators from defectors in these experiments, but appear to separate those that are more reactive to the behaviour of others from those that act categorically.
BibTeX
@inproceedings{monteroporras2022fast,
  author = {Eladio Montero-Porras and Tom Lenaerts and Riccardo Gallotti and Jelena Grujić},
  title = {{Fast deliberation is related to unconditional behaviour in iterated Prisoners’ Dilemma experiments}},
  booktitle = {Scientific},
  year = {2022},
  doi = {10.1038/s41598-022-24849-4},
}
Figure