Cognitive Resources Allocation and Cognitive Spillovers

2021-02-15

2021-06-30

Performance in the Memory game (fraction of uncovered pairs at the subject-round level)

- Rate of passive choices (based on an indicator variable = 1 if a subject follows the default in the Math exercise in a given round).

- Attention spans, measured by the total number of seconds devoted to a task in a given round.

- Performance in the Math exercises (based on an indicator variable = 1 if a subject solves the Math exercise correctly in a given per round)

- Total payoff (sum of earnings for the two tasks, individual-level averages across 9 rounds).

The experiment consists of two treatments, as described below. The experiment is played online; subjects are randomly assigned to one of the treatment conditions and keep their treatment assignment throughout the experiment. The experiment consists of 9 rounds; each round of the experiment lasts 60 seconds. In each round of the experiment, subjects can work on two tasks, a Memory game and a Math exercise.

The Memory game (also known as Concentration or Matching Pairs) consists of 10 pairs of cards with two-digit numbers on them. At the beginning of a round, the cards lie face down and can be flipped by a mouse click. Per move, subjects can flip two cards. If they find a pair - i.e., both cards show the same number - the cards stay flipped over. The objective of the task is to find as many pairs as possible. As Math exercise, subjects face three summations, each of which consists of six addends. Their task is to decide which of the three options yields the highest sum. The Math exercise features a default option that is implemented if subjects do not make an active decision. In particular, in each round, one option is randomly selected and displayed as the default choice.

Subjects see one task at a time on their screen and can freely navigate between the two tasks during the duration of each round. They navigate between the two tasks by using tabs at the top of their screen or buttons at the bottom of their screen. At the end of a given round, subjects face a waiting screen, on which they can enter the next round by clicking a button on the screen. Subjects face a new Memory game and Math exercise in each round.

Subjects receive EUR 0.04 for each uncovered pair in the Memory game; they receive EUR 0.20 if they select the option that yields the highest sum in the Math exercise in a given round. All rounds of the experiment are paid out to subjects. Subjects receive feedback on their earnings only after the final round of the experiment.

Treatments vary how tasks are displayed to subjects. When entering a new round in the baseline treatment, the Memory game is displayed to subjects. When entering a new round in the treatment condition, the Math exercise is displayed to subjects. In both treatments, they can freely navigate between tasks by using the on-screen Tabs / Buttons, as described above.

randomization done in office by a computer

individual

No

600 individuals

600 individuals, 9 rounds per individual

We aim at n=300 participants in each of the treatment cells.

We calibrate the sample size such that the minimal detectable effect size when comparing subjects’ performance in the Memory game across treatments is approximately 5 percentage points. For the power calculation, we assume that the fraction of uncovered pairs in the Memory game in the baseline treatment is 0.56 and the standard deviation is 0.24. Both of these numbers are informed by a pilot study. Assuming 300 observations per treatment cell and a power of 0.8, we are able to detect a performance difference of 5.4 percentage points across treatments.

This article investigates how incentives and behavioural policy interventions affect individuals’ allocation of scarce cognitive resources. Based on experimental evidence, we demonstrate that incentives systematically influence individuals’ allocation of cognitive resources, and their propensity to actively engage with a decision or to stay passive. Policies that steer individuals’ attention to a specific decision lead to more active decision-making and better choices in the targeted choice domain, but induce negative cognitive spillovers on the quality of choices in other domains. In our setting, these two countervailing effects offset each other, such that the overall payoff consequences of the interventions are essentially zero. We further document that cognitive spillovers are especially pronounced for complex choices and for subgroups of the population with a smaller stock of cognitive resources. We discuss implications for the design and evaluation of behavioural policy interventions.

Steffen Altmann, Andreas Grunewald, Jonas Radbruch, Interventions and Cognitive Spillovers, The Review of Economic Studies, Volume 89, Issue 5, October 2022, Pages 2293–2328.

https://doi.org/10.1093/restud/rdab087