Behavioral and neural consequences of orienting internal attention between short- and long-term memory
Poster Presentation 23.309: Saturday, May 16, 2026, 8:30 am – 12:30 pm, Banyan Breezeway
Session: Visual Working Memory: Interactions with long-term memory
Schedule of Events | Search Abstracts | Symposia | Talk Sessions | Poster Sessions
William Narhi-Martinez1, Kaiki Chiu1, Anna C. Nobre1; 1Yale University
We investigated how internal attention is directed to contents within or between short-term (STM) and long-term memory (LTM) across three experiments. Many contexts in daily life are associated with both STM and LTM content. We wanted to understand how attention operates when directed to one versus both of these internal domains. For LTM, participants learned the identities and locations of two objects within various scenes. The next day, the same scenes appeared with two new objects superimposed, which participants encoded into STM. At the end of the trial, participants made a decision about an item associated with the scene, either in STM or LTM. They selected which of two items was associated with the scene (behavioral Experiments 1 and 2) or decided whether a single item was associated with the scene (fMRI Experiment 3). Before the response array, a spatial retrocue could indicate which side the relevant item for the response had been located within the scene. The target item would come from either STM or LTM. Experiment 1 revealed significant performance benefits of internal spatial attention for valid vs. neutral retrocues for both STM and LTM targets. Experiment 2 showed that performance for LTM targets, but not for STM targets, was impaired when retrocues prioritized items from both STM and LTM domains compared to when contents from a single domain alone were prioritized. Initial results from Experiment 3 indicated dorsal frontoparietal regions exhibited significant clusters of activation specific to when participants were cued to combined STM and LTM contents, compared to when they were cued to contents purely in the STM or LTM domains. These findings challenge the standard view that once an LTM is recalled into working memory, it is maintained in an equivalent format to new STM information encoded from the external sensory stream.