A rational reinterpretation of dual-process theories

S. Milli, F. Lieder, & T. L. Griffiths

Cognition

Volume 217, December 2021, 104881

Abstract

Highly influential “dual-process” accounts of human cognition postulate the coexistence of a slow accurate system with a fast error-prone system. But why would there be just two systems rather than, say, one or 93? Here, we argue that a dual-process architecture might reflect a rational tradeoff between the cognitive flexibility afforded by multiple systems and the time and effort required to choose between them. We investigate what the optimal set and number of cognitive systems would be depending on the structure of the environment. We find that the optimal number of systems depends on the variability of the environment and the difficulty of deciding when which system should be used. Furthermore, we find that there is a plausible range of conditions under which it is optimal to be equipped with a fast system that performs no deliberation (“System 1”) and a slow system that achieves a higher expected accuracy through deliberation (“System 2”). Our findings thereby suggest a rational reinterpretation of dual-process theories.

From the General Discussion

While we have formulated the function of selecting between multiple cognitive systems as metareasoning, this does not mean that the mechanisms through which this function is realized have to involve any form

of reasoning. Rather, our analysis holds for all selection and arbitration mechanisms as having more cognitive systems incurs a higher cognitive cost. This also applies to model-free mechanisms that choose decision systems based on learned associations. This is because the more actions there are, the longer it takes for model-free reinforcement learning to converge to a good solution and the suboptimal choices during the learning phase can be costly.

The emerging connection between normative modeling and dual-process theories is remarkable because the findings from these approaches are often invoked to support opposite views on human (ir)rationality (Stanovich, 2011). In this debate, some authors (Ariely, 2009; Marcus, 2009) have interpreted the existence of a fast, error-prone cognitive system whose heuristics violate the rules of logic, probability theory, and expected utility theory as a sign of human irrationality.  By contrast, our analysis suggests that having a fast but fallible cognitive system in addition to a slow but accurate system might be the best

possible solution. This implies that the variability, fallibility, and inconsistency of human judgment that result from people’s switching between System 1 and System 2 should not be interpreted as evidence

for human irrationality, because it might reflect the rational use of limited cognitive resources. 

A pdf is here.

The Dynamics of Inattention in the (Baseball) Field

J. E. Archsmith, et al. 

IZA Institute of Labor Economics

June 2021

Abstract

Recent theoretical and empirical work characterizes attention as a limited resource that decision-makers strategically allocate. There has been less research on the dynamic interdependence of attention: how paying attention now may affect performance later. In this paper, we exploit high-frequency data on decision-making by Major League Baseball umpires to examine this. We find that umpires not only apply greater effort to higher stakes decisions, but also that effort applied to earlier decisions increases errors later. These findings are consistent with the umpire having a depletable ‘budget’ of attention. There is no such dynamic interdependence after breaks during the game (at the end of each inning) suggesting that even short rest periods can replenish attention budgets. We also find that an expectation of higher stakes future decisions leads to reduced attention to current decisions, consistent with forward-looking behavior by umpires aware of attention scarcity.

Conclusions

Conventional economic models embody agents able to make perfect, optimising decisions.  An important strand of recent efforts to increase the behavioral realism of models has been to acknowledge that attention is not costless---the effort required to attend to decisions and execute them well can be costly and cognitively tiring---and incorporate that in models. Models of “strategic inattention”, predicated on rational agents adjusting their behavior to account for attention being either limited and/or costly, are increasingly mainstream (for examples Caplin and Dean, 2015; Sims, 2003; Falkinger, 2011).

While the idea of costly attention is intuitively appealing, rigorous evidence characterizing its implications in real settings remains limited and primarily focuses on static effects in cross sectional data. This paper adds to and extends this evidence. Studying the quality of decisions made by a panel of professional decision-makers with strong incentives to get these decisions right, we show that MLB umpires systematically vary the effort they apply to individual decisions: applying greater attention to those associated with higher stakes. This is consistent with established theoretical models of strategic inattention. Our data-rich setting, in which the same umpire is called upon to issue a long series of decisions, allows for careful study of the dynamics of inattention and delivers our most novel results.

The paper is here.