Abstract
Memories are organized by meaning. For example, when recalling favorite restaurants, one might organize responses by the type of food served. Experimentally, this organization of associated experiences is captured by the free recall paradigm, in which participants have the tendency to consecutively recall stimuli that share meaning – a phenomenon termed semantic clustering. Intuitively, it would seem that semantic clustering is the result of processes during test – that is, a single retrieval cue may trigger the recall of multiple associated experiences. However, previous work has shown that study-phase neural signals predict subsequent semantic cluster- ing, suggesting that semantic organization is at least partly driven by study-phase processes, although it remains unclear what processes underlie these signals. Here, we hypothesized that semantic organization is supported by differential processing of associated study items, whereby processing of later study list items is influenced by the presentation of semantically associated items earlier in the study list. To test this hypothesis, we recorded scalp electroencephalographic (EEG) activity while participants performed a free recall task. Each study list was comprised of “pairs” of individually presented words, half of which were strongly semantically associated (e.g. “dog” and “cat”) and half of which were weakly semantically associated (e.g. “road” and “shore”). We separately assessed memory performance for first (“dog,” “road”) and second associates (“cat,” “shore”) and find that probability of recall is greater for strong second associates (“cat”) compared to weak second associates (“shore”). We find that second strong associates are more likely to be semantically clustered than first strong associates. We replicate the finding that semantic clustering is characterized by power increases over frontal and parietal/occipital brain regions. Critically, we show that the neural subsequent semantic clustering effect differs across pairs of semantic associates, with increased high frequency activity and decreased low frequency activity for second compared to first associates. Collectively, these results suggest that semantic organization of memories arises from differential processing of associated study items.
