Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

The influence of blocked versus interleaved training regimes and sleep on multi-task learning

Mina Habibi¹, Mehdi Senoussi², Pieter Verbeke, Senne Braem¹; ¹Universiteit Gent, ²Université Toulouse Jean Jaurès (Toulouse II)

Presenter: Mina Habibi

Recent studies suggest that humans benefit from blocked training in continual learning by promoting more factorized task representations. We investigated whether interleaved training may support human continual learning more when task separation is also aided by consolidation during sleep. Participants learned three tasks under blocked versus interleaved regimes across two experiments: Experiment 1 (contextual cues presented before stimuli; semantically non-informative labels) and Experiment 2 (stimuli presented before contextual cues; semantically informative labels). Testing occurred both immediately and after 24 hours. People trained under the blocked training regime showed higher accuracy during learning, but this advantage did not persist in the test phase of Experiment 1. In Experiment 2, blocked training resulted in higher accuracy across learning and testing. Critically, no sleep related benefits were found in either training regime for both experiments. RNNs fit to human data, however, revealed increased task separation from Day 1 to Day 2 in Experiment 2 across both training regimes. Our findings suggest that humans can benefit from both training regimes, and that the order and way in which context and stimulus are presented—rather than the regime alone—may play an important role in continual learning.

Topic Area: Memory, Spatial Cognition & Skill Learning

Extended Abstract: Full Text PDF