Memory and Anticipation
How to cite
Roy-Chowdhury, Vivek. 2024. "Memory and Anticipation." AEA RCT Registry. AEARCTR-0013581. https://doi.org/10.1257/rct.13581-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
We explore the role of memory in maintaining optimistic beliefs about the payoff consequences of past choices, consistent with models of anticipatory utility and optimal expectations. We do so in a setting designed to generate anticipatory motives: participants never learn the true consequences of their choices and can therefore distort their beliefs indefinitely. Relative to previous work, we use a design where the causal impact of the payoff-relevance of information is clearly identified: across treatment conditions, we manipulate the relationship between a signal and payoffs without affecting the objective distribution of the signal.
Experimental design
The experiment is split into two sessions, taking place a week apart. Subjects are randomised into two equally sized groups, Treatment and Control. Anticipatory motives to distort beliefs only exist in Treatment: the signal x is generated by the same objective distribution in both treatments but is only relevant to payoffs in Treatment. Since we are interested in the role of anticipatory motives, the challenge is to design an object of beliefs that may be relevant to utility but could remain uncertain indefinitely, maximising the potential for anticipatory motives. We opt to design x such that its only relevance is for to the consequences of subjects' choices for payoffs to a third party they are encouraged to care about.
The essential point is that at the beginning of session 1, subjects in Treatment choose whether to complete a task in session 2 to potentially increase the payment made to an assigned partner in Control. To introduce a guilt motive as well as a possible fairness one, they are informed that their partner will have to complete mandatory tasks in session 2 and receive a lower payment by default, as well as finding out what the subject chose and whether a payment was (or could have been) made.
The full belief treatment is only implemented for those who choose to do the tasks. It is revealed to these subjects that x is directly proportional to the probability p that the payment was made. Since we also measure beliefs on x for other participants, where x has no payoff-relevance (but belief accuracy is incentivised in the same way), our key analysis is of whether the payoff-relevance of x in Treatment causes beliefs to be initially distorted, and whether this distortion changes over time due to memory.
Design details
Cited by
No citation data available for this registration yet.