Select the next adaptive trial from candidate stimuli — select_gazepoint_adaptive

Provides a lightweight Bayesian-optimization-style acquisition helper for adaptive testing. It assumes candidate-level posterior means and standard deviations are already available or supplied by the user.

Usage

select_gazepoint_adaptive_trial(
  candidates,
  mean,
  sd,
  acquisition = c("ucb", "uncertainty", "expected_improvement"),
  kappa = 2,
  best_observed = NULL,
  maximize = TRUE
)

Arguments

candidates: A data frame of candidate stimuli/trials.
mean: Column containing posterior mean utility or expected information.
sd: Column containing posterior uncertainty.
acquisition: Acquisition rule: "ucb", "uncertainty", or "expected_improvement".
kappa: Exploration weight for UCB.
best_observed: Best observed value for expected improvement.
maximize: Logical; select maximum acquisition value if TRUE.

Value

One-row data frame corresponding to the selected candidate, with an added acquisition score.