Simulating the Scaling of Long-Term Memory Retrieval
Susanne Haridi, Mirko Thalmann, Eric Schulz, Max-Planck Institute for Biological Cybernetics, Germany
Session:
Posters 2B Poster
Presentation Time:
Fri, 25 Aug, 13:00 - 15:00 United Kingdom Time
Abstract:
The average human lives about 70 years, during which a growing number of memories are accumulated. But how are we still able to quickly retrieve currently relevant memories? We explore how the number of items (i.e., set size) affects retrieval time (RT), treating retrieval as a search process. Our study simulates the RTs using a sample-based retrieval process inspired by the search of associative memory model (Raaijmakers & Shiffrin, 1981). Our simulations show: 1. RTs increase with set size. 2. improved memory cues mitigate this effect and may alleviate retrieval failures. Specifically, increasing the similarity between cue and target improved recall performance, especially for larger set sizes. 3. the specificity of the cue was crucial for improving performance, and unspecific cues even increased RTs compared to baseline. These results suggest that effective memory retrieval crucially depends on the quality of the cue.