Prior expectation shapes the emotional response to sounds: behavioural and neural correlates

This study demonstrates that subjective prior expectations, rather than objective probabilities, directly bias the perceived aversiveness of salient sounds and modulate associated neural responses (P3, LPP, and alpha-beta oscillations), highlighting the critical role of top-down processing in shaping emotional perception.

The Gist

The Big Idea: Your Brain is a "Guessing Machine"

Imagine your brain isn't just a camera that passively records the world. Instead, it's a super-advanced weather forecaster. It constantly makes predictions about what's going to happen next based on what it has seen before.

Usually, when the weather forecast is right, you don't notice. But when the forecast is wrong (or when you aren't sure what the forecast is), your brain has to work overtime to figure out what's actually happening.

This study asked a specific question: Does knowing what's coming change how scary or annoying a sound feels?

The Experiment: The "Sound Roulette"

The researchers set up a game for 25 people using three types of visual clues (cues) on a computer screen:

1. The "X" Clue: This always meant a scary, scratchy sound (like a knife on a bottle) was coming next. (100% certain).
2. The "Y" Clue: This always meant a calm, water sound was coming next. (100% certain).
3. The "?" Clue: This was a mystery. It meant there was a 50/50 chance of either the scary sound or the calm sound. (Uncertain).

The Catch: Before the sound played, the participants had to guess, "How likely is it that I'm about to hear something scary?" After the sound played, they rated how much they hated it.

The Surprising Findings

1. The "Group Average" Lie

If you just looked at the average rating of everyone in the room, the results were boring. It seemed like knowing what was coming didn't really change how much people hated the scary sounds. Whether they knew it was coming or were surprised, they still hated it.

The Metaphor: Imagine a room full of people eating a very spicy pepper. If you ask the whole room, "Did knowing it was spicy change how hot it tasted?" the average answer might be "No."

2. The "Individual Truth"

However, when the researchers looked at each person individually, a different story emerged.

• If a person really expected the scary sound, they rated it as more annoying.
• If a person didn't expect it, they rated it as slightly less annoying.

The Metaphor: It turns out that for some people, the "spicy pepper" tasted hotter if they were warned it was coming. Their brain was already bracing for impact, so when the pepper hit, it felt like a double dose. The study found that your personal expectation acts like a volume knob for your emotions. If you expect the noise, your brain turns the "annoyance volume" up.

3. The Brain's "Surprise Alarm" (EEG Results)

The researchers also hooked the participants up to EEG machines (which read brain waves) to see what was happening inside their heads.

• The "P3" and "LPP" Signals: These are like little lightning bolts in the brain that happen when something grabs your attention.
• The Result: When the scary sound happened after the mystery "?" cue, the brain's lightning bolts were huge. But when the sound happened after the predictable "X" cue, the lightning bolts were smaller.
• Why? When the brain is surprised by a bad sound, it has to scramble to process it. When it's expecting it, the brain is already prepared, so it doesn't need to scream as loud. Interestingly, this "surprise alarm" only went off for the scary sounds, not the calm water sounds.

4. The "Beta Wave" Brakes

The study also looked at brain waves called Beta waves. Think of these as the brain's brakes or focus gears.

• When people were very confident in their prediction (e.g., "I know for sure the scary sound is coming"), their brain's Beta waves slowed down (suppressed). This is like the brain saying, "I've got this under control; I'm ready."
• When people were uncertain, those brakes didn't engage as well.

The Takeaway: Why This Matters

This study teaches us two main things:

1. Expectation is Reality: Even for loud, obvious, annoying sounds, your brain doesn't just react to the sound itself. It reacts to what you thought was going to happen. If you expect the worst, you often experience the worst.
2. Don't Trust the Average: If you only look at the group average, you might miss the fact that expectations are powerful for individuals. Just because the "average person" didn't seem affected doesn't mean you won't be.

In a nutshell: Your brain is a storyteller. If the story says "Scary sound is coming," your brain writes the chapter where you feel terrified. If the story says "Maybe, maybe not," your brain writes a chapter where you are confused and startled. The sound is the same, but the story changes how you feel.

Technical Summary

1. Problem Statement and Research Gap

While Bayesian models of cognition suggest that perception is a precision-weighted combination of top-down expectations and bottom-up sensory input, this relationship is well-established for ambiguous stimuli but less clear for highly salient, affective stimuli.

• The Conflict: Affective stimuli (e.g., aversive sounds) are typically processed via dominant bottom-up mechanisms due to their high salience. It is unclear how top-down expectations compete with this strong sensory input to alter the perceived emotional valence.
• The Methodological Gap: Previous studies often rely on objective cue validity (e.g., a cue predicts a sound 100% of the time) as a proxy for subjective certainty. However, participants may not internally "feel" certain even when objective probability is high. This study argues that subjective expectation (individual reported confidence) is the critical variable, particularly when competing with salient stimuli.
• The Neural Gap: While ERP components like P3 and LPP are known to reflect emotional salience, their modulation by expectation in the context of aversive sounds remains inconsistent. Furthermore, the specific oscillatory mechanisms (e.g., alpha/beta bands) encoding the precision of predictions in affective processing are under-explored.

2. Methodology

Participants:

• 25 healthy adults recruited; 22 remained after excluding those who failed to learn cue-sound associations.
• All had normal hearing and no neurological/psychiatric history.

Stimuli and Paradigm:

• Sounds: 10 one-second sounds (5 aversive "scraping/scratching" sounds; 5 neutral "water" sounds).
• Cues: Three visual cues presented for 500ms:
  • 'X' (Certain Aversive): 100% probability of aversive sound.
  • 'Y' (Certain Neutral): 100% probability of neutral sound.
  • '?' (Uncertain): 50% probability of either sound.
• Trial Structure:
  1. Cue presentation.
  2. Subjective Expectancy Rating: Participants rated the likelihood (0–100) of an aversive sound following the cue.
  3. Anticipatory interval (4s).
  4. Sound presentation (1s).
  5. Aversiveness Rating: Participants rated the sound's aversiveness (0–10).
• Design: 80 trials per condition (certain aversive, certain neutral, uncertain) in a pseudorandomized order.

Neural Recording & Analysis:

• EEG: 64-channel BioSemi system.
• Preprocessing: Low-pass filtered (100 Hz), downsampled (256 Hz), detrended (no high-pass filter to avoid distortion), ICA for ocular artifact removal (AMICA), and RANSAC for bad channel interpolation.
• ERP Analysis: Focused on P3 (300ms post-sound) and Late Positive Potential (LPP) (400–1000ms post-sound).
• Time-Frequency (TF) Analysis: Complex Morlet wavelets (2–47 Hz). Baseline corrected pre-cue.
• Statistical Approach:
  • Mediation Analysis: Used to test if subjective expectancy mediates the relationship between cue type (certain/uncertain) and perceived aversiveness.
  • Uncertainty Index (UI): A novel metric calculated as $|Probability_{Reported} - Probability_{Real}|$ to quantify prediction precision/confidence.
  • Correlations: TF power correlated with UI and standard deviation of ratings to isolate precision effects from motor preparation.

3. Key Results

Behavioral Findings:

• Learning: Participants successfully learned the contingencies (Mean expectancy: Aversive cue=88.9, Neutral cue=12.5, Uncertain=52.9).
• Reaction Times: Participants were significantly slower (~200ms) during the uncertain condition, confirming induced cognitive uncertainty.
• Perceived Aversiveness (Group Level): A 2x2 ANOVA showed no significant main effect of certainty on the group-level aversiveness ratings. The manipulation did not shift the group average.
• Perceived Aversiveness (Individual Level): Mediation analysis revealed a significant effect. Subjective expectancy acted as a full mediator:
  • Higher expectation of an aversive sound $\rightarrow$ Higher reported aversiveness.
  • Higher expectation of a neutral sound $\rightarrow$ Lower reported aversiveness.
  • Conclusion: Expectation biases perception, but this effect is masked at the group level due to inter-subject variability in prediction precision.

Neural Findings (ERPs):

• P3 Component: Uncertain (unpredicted) aversive sounds elicited a significantly larger P3 amplitude compared to certain aversive sounds. This difference was absent for neutral sounds.
• LPP Component: The early-LPP (400–600ms) was significantly enhanced for uncertain aversive sounds compared to certain ones. Later LPP windows showed no significant differences.
• Interpretation: Top-down expectation modulates neural processing of salient stimuli, specifically amplifying surprise signals (P3/LPP) when the prediction is violated or uncertain.

Neural Findings (Time-Frequency/Oscillations):

• Cue Processing (Pre-stimulus):
  • Alpha-Beta Suppression: Certain cues (high precision) induced stronger alpha-beta (approx. 15–34 Hz) power suppression compared to uncertain cues.
  • Correlation with Precision: There was a strong positive correlation between the Uncertainty Index (UI) and beta power. Lower uncertainty (higher precision) = Greater beta suppression.
  • Control: These correlations were not driven by motor preparation (no correlation with reaction times).
• Sound Processing (Post-stimulus):
  • Beta Suppression: Unpredicted (uncertain) aversive sounds elicited stronger beta power suppression (~400ms post-onset) compared to predicted sounds.
  • Theta Dissociation: Uncertain aversive sounds showed theta decrease, while uncertain neutral sounds showed theta increase.
  • Interpretation: Beta suppression appears to index both the precision of the prior prediction (during the cue) and the prediction error/updating process (during the sound).

4. Key Contributions

1. Subjective vs. Objective Expectation: The study demonstrates that subjective expectation is the critical driver of perceptual bias in affective processing. Relying solely on objective cue validity (100% vs 50%) fails to capture the individual variability in how priors influence the perception of salient stimuli.
2. Mediation of Affective Perception: It provides empirical evidence that expectations act as a mediator between environmental cues and the subjective experience of aversiveness, even for highly salient, non-ambiguous sounds.
3. Neural Mechanisms of Precision:
   • Identifies alpha-beta oscillations as a neural signature of prediction precision. Stronger suppression indicates higher confidence/precision.
   • Shows that P3 and LPP components are specifically sensitive to the violation of expectations regarding aversive (but not neutral) stimuli, suggesting a specialized neural pathway for unexpected negative affect (potentially involving the insula/amygdala).
4. Methodological Rigor: The use of mediation analysis and the "Uncertainty Index" allows for a more granular understanding of the Bayesian weighting process in individual subjects.

5. Significance and Implications

• Bayesian Cognition: The results support the Bayesian framework, showing that even for high-salience stimuli, top-down priors compete with bottom-up input. The "weight" of the prior varies by individual precision.
• Clinical Relevance: Understanding how expectation modulates aversive processing is crucial for conditions involving anxiety, phobias, or chronic pain, where maladaptive priors (e.g., expecting pain) may amplify the subjective experience of negative stimuli.
• Neural Biomarkers: The study proposes alpha-beta suppression as a potential biomarker for "prediction precision" and the P3/LPP enhancement as a marker for "surprise" in affective contexts.
• Limitations: The authors note that the use of a question mark ('?') as an uncertain cue may have carried semantic meaning (signaling "unknown") distinct from a purely random cue, which could influence attentional mechanisms differently than associative learning.

In summary, this paper establishes that expectations directly alter the perception of affective stimuli, a process mediated by individual prediction confidence and indexed by specific oscillatory (alpha-beta) and ERP (P3/LPP) signatures.