The Effects of Learnability and Reward Responsiveness on Reward Processing

Although task learnability did not generally alter reward positivity (RewP) amplitudes, restricting the analysis to high performers revealed that individuals with low reward responsiveness exhibited enhanced RewP specifically in learnable tasks, highlighting a critical interaction between contextual factors and individual differences in reward processing.

The Gist

The Big Question: Does "Figuring It Out" Make Rewards Feel Better?

Imagine you are playing a video game.

• Scenario A: You press buttons randomly. Sometimes you get a coin, sometimes you don't. You have no idea why. It's pure luck.
• Scenario B: You notice that pressing the "Jump" button on a red platform always gives you a coin, but on a blue platform, it never does. You figure out the pattern.

The researchers wanted to know: Does your brain react differently to winning a coin in Scenario B (where you learned the rules) compared to Scenario A (where it's just random)?

They also wondered if this depends on your personality. Some people are naturally super excited about rewards (like a dog seeing a treat), while others are a bit more chill. Does the "learning" part matter more for the excited people or the chill people?

The Experiment: The "Doors" Game

The researchers put 38 people in a lab and had them play a computer game called the "Doors Task."

• The Setup: Two doors appear on the screen. You pick one.
• The Twist: They played two types of rounds:
  1. The "Learnable" Rounds: One door was rigged to give you money 60% of the time, and the other only 10% of the time. If you paid attention, you could figure out which door was the "good" one.
  2. The "Unlearnable" Rounds: The computer shuffled the results randomly. It didn't matter which door you picked; it was pure chance.

While they played, the researchers hooked them up to an EEG cap (a brain-sensing helmet) to measure a specific electrical signal called the RewP. Think of the RewP as a "Happy Spark" in the brain that happens milliseconds after you win something. Scientists use this spark to measure how much your brain cares about rewards.

What They Found

1. The "Fun" Factor (Behavioral Results)

The Result: People loved the "Learnable" rounds more. They reported having more fun, feeling more motivated, and feeling like they were doing a better job.
The Analogy: Imagine eating a meal where you have to guess the ingredients. In the "Unlearnable" version, it's just random flavors. In the "Learnable" version, you start to realize, "Hey, if I add salt, it tastes great!" Even if the food is the same, the act of figuring it out makes the experience more enjoyable.

2. The "Happy Spark" (Brain Results)

The Result: Surprisingly, the "Happy Spark" (RewP) in the brain was exactly the same size whether the game was learnable or unlearnable.
The Analogy: You might expect that figuring out the secret code would make the brain's "winning" signal explode with joy. But instead, the brain's spark was the same size whether you were guessing blindly or playing with a cheat sheet. The researchers were initially confused because they thought learning would make the spark bigger.

3. The Personality Twist (The Hidden Discovery)

The Result: When they looked closer, they found a secret pattern. The relationship between "learning" and the "Happy Spark" depended on how well the person played the game and how sensitive they were to rewards.

• The High Performers: People who were good at the game and naturally sensitive to rewards showed a bigger "Happy Spark" when they were in the learnable mode.
• The Low Performers: People who struggled to figure out the pattern (or weren't naturally reward-sensitive) actually showed the opposite effect.

The Analogy: Imagine a group of people trying to solve a maze.

• The smart, motivated explorers get a huge adrenaline rush when they finally find the path (the learnable condition).
• The confused wanderers get frustrated and their brain signal actually drops when they realize the maze is tricky, or they just give up and treat it like a random walk.

Why Does This Matter?

This study is important for two main reasons:

1. It challenges how we study the brain: Most studies use the "Unlearnable" (random) version of this game because it's easier to control. But this paper suggests that if you only use random games, you might be missing how the brain reacts when people feel like they are actually learning and in control.
2. It helps understand mental health: The "Happy Spark" (RewP) is often used as a test for depression. People with depression often have a "dimmer" spark. This study suggests that maybe the spark isn't just about being depressed; it might also depend on whether the person feels like they can learn and succeed in the task. If a task feels impossible or random, even a healthy brain might show a smaller spark.

The Takeaway

Learning to win feels good, but it doesn't always make your brain's "winning signal" bigger for everyone.

It turns out that how your brain reacts to a reward depends on a complex mix of:

1. Whether you feel like you can learn the rules.
2. How good you are at the game.
3. Your natural personality regarding rewards.

So, the next time you feel frustrated playing a game, remember: your brain might be reacting not just to the loss, but to the feeling that the game is unlearnable!

Technical Summary

1. Problem Statement and Research Gap

Reward processing is frequently studied using the "doors" task, a two-choice paradigm where outcomes are random and unlearnable. This design is methodologically preferred to ensure equal probability of gains and losses, avoiding confounds with probability-sensitive components like the P300. However, this creates a theoretical disconnect:

• Theoretical Conflict: Reinforcement learning theories posit that the Reward Positivity (RewP), an event-related potential (ERP) component reflecting dopaminergic prediction errors, is a learning signal driven by the updating of action-outcome associations.
• The Gap: If the standard task is unlearnable, reliable predictions cannot form. It remains unclear whether the RewP is sensitive to task learnability (the ability to acquire action-outcome contingencies) and whether this sensitivity is moderated by individual differences in reward responsiveness (specifically the Behavioral Activation System, or BAS).
• Clinical Relevance: The RewP is a proposed biomarker for internalizing disorders (e.g., depression). If the RewP is heavily influenced by whether a task is perceived as learnable, current clinical interpretations using unlearnable tasks may be incomplete or confounded by task engagement levels.

2. Methodology

Participants

• Sample: 38 undergraduate students (final N after excluding 2 for excessive EEG artifacts) from MacEwan University.
• Demographics: Mean age = 21.58 years; 13 males.
• Screening: No neurological impairments; normal/corrected vision.

Experimental Design

• Task: A modified "doors" task consisting of 20 blocks of 20 trials each.
• Conditions:
  • Learnable Blocks (10 blocks): One door had a 60% reward probability, the other 10%. The high-probability door was randomized per block.
  • Unlearnable Blocks (10 blocks): Outcomes were random and unlinked to choices. Crucially, the sequence of outcomes was yoked to the learnable blocks to ensure matched overall reward frequencies (preventing probability confounds).
• Procedure:
  • Participants completed the BIS/BAS scale (Carver & White, 1994) to measure individual differences in reward responsiveness.
  • EEG was recorded during the task.
  • Post-block surveys assessed contingency awareness ("Which door was better?") and subjective engagement (fun, motivation, perceived performance).

Data Acquisition and Analysis

• EEG: 32-channel cap, sampled at 1000 Hz, re-referenced to mastoids.
• Preprocessing: Bandpass filtered (0.1–30 Hz), ICA for ocular artifact removal, epoching (-200 to 600 ms relative to feedback).
• ERP Metric: The RewP was quantified as the mean voltage difference (Win - Loss) at electrode FCz within the 240–340 ms window.
• Statistical Approach:
  • Paired t-tests for behavioral engagement and main ERP effects.
  • Correlation analysis between BAS scores and the learnability-related RewP difference score ($\Delta$RewP = Learnable - Unlearnable).
  • Exploratory Analysis: Multiple regression testing the interaction between Reward Responsiveness (BAS) and Task Performance (indexed by win-stay probability) on $\Delta$RewP.

3. Key Results

Behavioral Findings

• Learning: Participants successfully learned the contingencies in the learnable blocks, with performance increasing above chance levels over trials.
• Engagement: Participants reported significantly higher levels of fun, motivation, and perceived performance in the learnable condition compared to the unlearnable condition, despite matched outcome frequencies.

ERP Findings (Main Effects)

• RewP Presence: A robust RewP (more positive for wins than losses) was observed in both conditions.
• Learnability Effect: Contrary to predictions, there was no significant main effect of task learnability on RewP amplitude ($p = .56$). The RewP did not differ between learnable and unlearnable blocks at the group level.
• Moderation by BAS: There was no significant correlation between BAS reward responsiveness scores and the $\Delta$RewP ($r = 0.04, p = .81$).

Exploratory Findings (Interaction Effects)

• A significant interaction was found between Reward Responsiveness and Task Performance (win-stay probability) in predicting $\Delta$RewP ($p = .004$).
• High Performers: Among participants who performed well (high win-stay probability), those low in reward responsiveness showed an enhanced RewP in the learnable condition compared to the unlearnable condition.
• Low Performers: Participants with low performance showed the opposite pattern (or no clear effect), suggesting they may not have detected the underlying task structure.

4. Key Contributions

1. Contextual Nuance in Reward Processing: The study demonstrates that while the RewP is robustly present even in unlearnable (random) tasks, the modulation of this signal by task context is not uniform. It depends on the participant's ability to detect and utilize task structure.
2. Interaction of State and Trait: The findings reveal a complex interplay where individual differences (BAS traits) only manifest in neural reward signals when the individual is behaviorally engaged enough to perceive the task as learnable.
3. Refinement of the RewP Biomarker: The results suggest that "blunted" RewP responses in clinical populations might not solely reflect a deficit in reward sensitivity. Instead, they could be influenced by task engagement and the perception of control/learnability. If a patient cannot detect contingencies (due to cognitive deficits or disengagement), the RewP may be attenuated regardless of their underlying reward sensitivity.

5. Significance and Implications

• Methodological Implications: Researchers using the doors task to measure the RewP must consider that the "unlearnable" nature of the task may suppress specific learning-related neural dynamics in certain subgroups. Future studies should account for individual differences in strategy use and performance.
• Clinical Implications: The RewP's utility as a biomarker for depression and other disorders requires careful interpretation. A blunted RewP might reflect a failure to engage with the task structure (learnability) rather than a fundamental deficit in the reward system.
• Theoretical Implications: The study supports the view that reward processing is an integrated evaluation of individual traits (BAS), contextual factors (learnability), and behavioral strategy (performance). It suggests that the neural teaching signal (RewP) is most sensitive to learning when the participant is both motivated (low BAS) and capable of detecting the learning structure (high performance).

Conclusion:
While task learnability did not produce a broad group-level enhancement of the RewP, it significantly interacted with individual differences in reward responsiveness and task performance. Specifically, low-reward-responsive individuals showed enhanced neural reward sensitivity only when they were high-performing in a learnable environment. This highlights the necessity of considering task engagement and performance when interpreting the RewP in both basic and clinical neuroscience.