We test whether LLMs show robust decision biases. Treating models as participants in two-arm bandits, we ran 20000 trials per condition across four decoding configurations. Under symmetric rewards, models amplified positional order into stubborn one-arm policies. Under asymmetric rewards, they exploited rigidly yet underperformed an oracle and rarely re-checked. The observed patterns were consistent across manipulations of temperature and top-p, with top-k held at the provider default, indicating that the qualitative behaviours are robust to the two decoding knobs typically available to practitioners. Crucially, moving beyond descriptive metrics to computational modelling, a hierarchical Rescorla-Wagner-softmax fit revealed the underlying strategies: low learning rates and very high inverse temperatures, which together explain both noise-to-bias amplification and rigid exploitation. These results position minimal bandits as a tractable probe of LLM decision tendencies and motivate hypotheses about how such biases could shape human-AI interaction.
Figures & tables
Strategy
Temperature
Top- p
Strict
0.0
0.5
Moderate
1.0
0.5
Default-like
1.0
1.0
Exploratory
2.0
1.0
Table 1: Decoding Configurations with Temperature and Top- p Settings
Metric
Definition
Total Reward
Sum over trials [ 22 ]
Target-arm Rate
Fraction choosing the higher-probability arm [ 22 ]
Loss-Shift Win-Shift
Probability of switching after a loss / win [ 22 ]
Choice Bias Index
cˉ−0.5,cˉ=P(choose Y) [ 15 ]
Stubbornness Rate
Fraction of runs with cˉ≥0.8 or cˉ≤0.2 [ 12 ]
Amplification Index
Fraction of post-warm-up runs that are monomorphic [ 9 ]
Table 2: Summary of Computed Behavioural Metrics
Figure 1: Symmetric Bandit Behavioural Metrics
Figure 2: Asymmetric Bandit Behavioural Metrics
Figure 3: Posterior Densities of Group - level Parameters
Figure 4: Test-retest Reliability on the Symmetric Bandit
Figure 5: Test-retest Reliability on the Asymmetric Bandit