Does a bad mood change how we predict words while reading?
After watching sad film clips, readers stopped using verb meaning to predict who a sentence would talk about next, while their grammar processing stayed largely intact.
Source
How robust is the language architecture? The case of mood
Study at a glance
- Design
- Human experiment — Within-subject EEG/ERP study with positive (Happy Feet) versus negative (Sophie's Choice) film-clip mood induction in two sessions a week apart, measuring ERPs to pronouns that fit or violated verb-based implicit causality and to subject-verb agreement errors; plus a between-subjects story-completion experiment
- N
- N=32 · 55 women recorded for EEG, 32 analysed after exclusions (16 in the 'mood-responsive' subgroup); 40 different women in the behavioural story-completion study (20 per mood condition)
- Population
- Right-handed female native Dutch speakers, mostly university-aged, without neurological or reading impairment
- Outcome
- ERP positivity (400-700 ms) to bias-inconsistent pronouns, P600 to agreement violations, N1/P2; self-reported mood; proportion of bias-consistent story completions
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Women read short Dutch stories word by word while their brain activity was recorded with EEG, in two sessions a week apart: one after cheerful clips from Happy Feet and one after sad clips from Sophie's Choice. Some stories used verbs like 'praise' or 'apologize' that make one person the likely cause, followed by a pronoun that either fitted or clashed with that expectation. Other stories contained grammar errors such as 'The boys turns' as a control. In a follow-up, 40 new women saw the same clips and simply completed the stories after 'because' to see whether mood changed the verb bias itself.
What they found
After the cheerful clips, pronouns that clashed with the verb's implied cause produced a positive brain-wave deflection around 400-600 ms, showing readers had been predicting the other person. After the sad clips this effect disappeared, and the difference between moods was significant; among the 16 participants whose mood clearly changed, the pattern was especially clear. Mood did not change early visual brain responses or the size of the P600 response to grammar errors. In the story-completion study, both mood groups gave about 90% bias-consistent completions, so a bad mood did not weaken the verb bias itself.
The limits
What it doesn't show
The film clips only worked as intended for about half the EEG participants, and the clearest results come from that self-selected subgroup of 16, which weakens the causal claim. Only women were tested, and the positive clips did not raise mood above baseline, so the study really compares a maintained mood with a worsened one. The two films also differ in complexity, realism and arousal, not just valence, so something other than mood could contribute. The finding concerns one type of prediction (implicit causality) in reading, and the authors do not show that all language anticipation is affected.
Key terms
- Implicit causality
- Knowledge attached to verbs like 'praise' or 'apologize' about which person is likely the cause of the event, shaping who we expect to be mentioned after 'because'.
- Event-related potential (ERP)
- An average EEG brain response time-locked to a specific event, such as reading a particular word.
- P600
- A positive ERP wave peaking around 600 ms after a word, typically elicited by grammatical violations or reanalysis.
- Mood induction
- An experimental technique, here watching film clips, used to temporarily shift participants' mood in a positive or negative direction.
- Heuristic processing
- Fast, rule-of-thumb thinking that relies on typical patterns rather than careful analysis.
- Narrative transportation
- The tendency to become absorbed in and feel along with characters in stories or films.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
After 'Linda praised David because...', who do readers typically expect to be mentioned next?
Common questions
Why include grammar errors in the study?
They served as a control: if a bad mood simply made people stop paying attention, grammar responses should shrink too. The P600 to agreement errors stayed about the same size, suggesting the mood effect was specific to heuristic prediction.
If a bad mood removes prediction, why did the story-completion study show no mood effect?
Completing a story gives people time to think, and they still used the verb bias about 90% of the time in both moods. The authors conclude a bad mood affects the rapid, on-the-fly use of the bias during reading, not the knowledge itself.
Why did the mood manipulation fail for some people?
Participants whose mood shifted tended to score higher on a scale measuring how much they feel along with fictional characters, and some others arrived in a better mood on the sad-film day, which counteracted the manipulation.
More on Language