The dataset is organized in a wide-format structure, where each row corresponds to a single participant and variables represent item-level responses recorded at each learning trial. Specifically, there is one column for each trial × item combination, such that for every trial t∈{1,…,K} and every item i∈{1,…,N}, the observed response is stored as a distinct variable. Column names follow a deterministic convention that explicitly encodes both indices (e.g., T{t}_I{i}), enabling unambiguous reconstruction of the person × (trial×item) response matrix directly from the header without requiring additional indexing variables at the record level. This layout supports straightforward extraction of trial-specific blocks (all items within a trial) or item-specific series (the same item tracked across trials), facilitating reproducible data processing and verification. Each trial-by-item variable is conceptually defined by three components: trial, which denotes the ordinal learning trial within the acquisition phase; item, which identifies the target stimulus; and score, which encodes the participant’s performance on that item at that trial. The score is stored as a binary indicator with the following coding: 0 = incorrect / not recalled and 1 = correct / recalled.