Why Can Describing a Face Make It Harder to Recognise?

We usually expect that putting an experience into words will strengthen memory. Faces can sometimes behave differently. After seeing an unfamiliar face, a person who immediately describes the eyebrows, nose, hair and facial shape in detail may later be less accurate when choosing that face from a set of photographs. This average pattern is called verbal overshadowing.

In the classic 1990 experiments, participants watched a simulated crime video. Some then described the perpetrator's appearance, while others completed an unrelated task. The description group performed worse in a later photo-line-up. Subsequent studies and a registered replication have supported the effect, but it does not occur in every experiment, every person or with every kind of face.

Its mechanism remains debated. One account holds that face recognition relies heavily on relationships within the whole configuration, whereas words draw attention towards a few nameable features. At test, a person may then rely too heavily on an incomplete verbal representation. Another account proposes that describing temporarily changes the style of retrieval without permanently rewriting the visual memory. Delay, description instructions, facial distinctiveness and test design can all change the result.

The lesson is not that talking destroys memory. Language may still help with the sequence of events, actions and clearly nameable details. Verbal overshadowing is a narrower claim: forcing a perceptual experience that is difficult to verbalise fully into a list of local features can make later recognition depend on unsuitable cues. The central point is not that vision is truer than language, but that memory performance depends on whether the way information was processed suits the later task.

https://pubmed.ncbi.nlm.nih.gov/9099066/
https://pubmed.ncbi.nlm.nih.gov/25987518/