Biasing effect of advocacy: Not controllable but automatic, even in highly stable beliefs, when people were financially incentivized to form true beliefs & among lawyers (trained to prevent advocacy from biasing their judgements)

The automatic influence of advocacy on lawyers and novices. David E. Melnikoff & Nina Strohminger. Nature Human Behaviour (2020). Sep 7 2020. https://www.nature.com/articles/s41562-020-00943-3

Abstract: It has long been known that advocating for a cause can alter the advocate’s beliefs. Yet a guiding assumption of many advocates is that the biasing effect of advocacy is controllable. Lawyers, for instance, are taught that they can retain unbiased beliefs while advocating for their clients and that they must do so to secure just outcomes. Across ten experiments (six preregistered; N = 3,104) we show that the biasing effect of advocacy is not controllable but automatic. Merely incentivizing people to advocate altered a range of beliefs about character, guilt and punishment. This bias appeared even in beliefs that are highly stable, when people were financially incentivized to form true beliefs and among professional lawyers, who are trained to prevent advocacy from biasing their judgements.

Discussion
It may be no coincidence that, when describing biased thinking, psychologists reach for the metaphor of the advocate32–34. Our findings run counter to Gorgias’s assertion that one can advocate for a cause while remaining impartial. Instead, they support the Socratic view that advocacy induces an automatic cognitive bias, one that inevitably distorts the beliefs of those who merely assume the role of advocate.

In addressing whether the biasing effect of advocacy is uncontrollable, our findings leave open the question of whether this effect is unintentional, efficient or unconscious. Though all of these processing features fall under the umbrella of automaticity, the presence or absence of one cannot be inferred from the presence or absence of another8,10,11. Accordingly, additional work is needed to determine whether the biasing effect of advocacy can occur in the absence of a goal to advocate (unintentionally), under cognitive load (efficiently) or without the advocate’s awareness (unconsciously). Controllability, however, stands out among the various processing features as the most important for advocacy. The question of whether advocates can resist bias sparked the debate between Gorgias and Socrates, and the belief that they can continues to inform legal systems around the world4–7. That this belief may well be wrong has major implications for the law and beyond.

Lawyers are trusted to give their clients guidance based on objective, unbiased assessments of the evidence. When they cannot remain impartial, their advice will be of lower quality. For instance, a defence attorney who underestimates the likelihood of a guilty verdict may be less likely to advise his or her client to accept a plea bargain. Besides lawyers, many scientists aim to marry advocacy and impartiality by using their own research to influence policy decisions. Entrepreneurs have a vested interest in simultaneously pursuing the interests of their shareholders while retaining a realistic view of their company’s prospects.

The possibility remains, of course, that a means to control the biasing effect of advocacy will yet be discovered. Only through further investigation can the probability of finding such a strategy be so reduced that the biasing effect of advocacy can be declared, with practical certainty, uncontrollable. Until then, decision makers are advised that this bias seems remarkably difficult to overcome, and a reliable strategy for doing so remains elusive. Optimizing advocacy may depend not on controlling biased thinking but on external systems and structures that keep such thinking from shaping decisions.


---
Check also Everybody argues and everybody wins: Overestimation of success as a driver of debate Logg, Jennifer M.; Berg, Logan A.; Minson, Julia A. Presented at the Society for Judgment and Decision Making, The 2019 40th Annual Conference. Montréal, Canada
November 15–18, 2019. http://www.sjdm.org/programs/2019-program.pdf

Abstract: We examine why people argue despite limited success in persuading others. In 10 experiments (N=2,911), our participants overclaim past argument success (Experiment 1) and mispredict future success (Experiments 2-6). Mispredicting future success results in higher entry into arguments and a greater willingness to bet money on predicted success (Experiment 2). Such biased assessments are robust to constrained definitions of winning (Experiments 3A-3D) and increases when the definition is based on logic (Experiment 4). Biased assessments of argument success seem to stem more from overconfidence in the "correctness" of individual's views than in debate skills (Experiment 5 & 6).