# The timescale and direction of influence of a third inferior alternative in human value-learning

## TL;DR

This study shows how the presence of a less valuable option affects how people learn and evaluate the value of other options, with effects that persist even when the inferior option is no longer available.

## Contribution

The study introduces a new mechanism for context-dependent valuation based on sequential binary comparisons, which better explains value modulation than existing normalization theories.

## Key findings

- Context-dependent valuation is evident in pre-choice value estimates and is consistent in both binary and ternary choices.
- The modulation of value representations is driven by temporal context, not immediate context.
- The effects of inferior options on value perception persist even when those options are no longer present in the choice set.

## Abstract

The way humans and other animals represent the values of alternatives is context-dependent, as it can be distorted by inferior alternatives that are immediately available for choice (immediate context); or that were encountered in previous episodes (temporal context). Yet, the extent to which the immediate and temporal context (co-) shape context-dependent valuation remains unclear. Here, we asked human participants (onsite: N = 30, online: N = 68) to learn the values associated with three alternatives and explicitly report these values before making binary and ternary choices among the alternatives. We show that context-dependent valuation is evident in the pre-choice value estimates and manifests equally in binary and ternary choices. Accordingly, we conclude that value representations are modulated by the temporal (and not the immediate) context. The direction and across-participants variability of this modulation cannot be captured by extant normalization theories but by a mechanism constructing values through sequential binary comparisons.

Increasing the value of an inferior option makes two superior ones look less valuable. This distortion emerges during learning and persists in later choices, even in choice sets that omit the inferior option.