# Learning about causal relations that change over time: primacy and recency over long timeframes in causal judgments and memory

## TL;DR

People can detect changes in cause-effect relationships over time, showing better performance in short-term tasks and recency effects in long-term tasks.

## Contribution

The study reveals how people track changing causal relations and memory patterns over long timeframes, challenging existing theories.

## Key findings

- Participants detected changes in cause-effect relations better in short tasks but still succeeded in long-term tasks.
- Recency effects were observed in long-term judgments, while short-term judgments lacked primacy or recency effects.
- Episodic memory was poor, but primacy and recency effects were present in short-term tasks and weakened over time.

## Abstract

Being able to notice that a cause–effect relation is getting stronger or weaker is important for adapting to one’s environment and deciding how to use the cause in the future. We conducted an experiment in which participants learned about a cause–effect relation that either got stronger or weaker over time. The experiment was conducted with a typical procedure in which the learning cases were presented rapidly, and with a mobile phone procedure, in which participants experienced the cause–effect relation over 24 days. First, we found that people could detect the change in contingency. They were better at doing so in the artificial short timeframe task, but still could do so in the more realistic long timeframe task. Second, when making summary judgments about the cause–effect relation, participants exhibited a recency effect for most measures in the long timeframe, but did not exhibit a primacy or recency effect in the short timeframe. Third, though participants’ episodic memories for individual cause–effect events in the learning sequence were quite poor, they did exhibit primacy and recency effects in the short timeframe; these were attenuated in the long timeframe. These findings raise fundamental questions about causal learning; they suggest that people automatically recognize changes and store representations of the contingency during different phases of learning, but this ability is not predicted by most existing theories of causal learning.

The online version contains supplementary material available at 10.1186/s41235-025-00614-9.