Despite decades of research on the mechanisms of memory in humans, several questions regarding the spatial and temporal correlates of sensory memory remain unanswered. We have been working on memory for meaningless auditory stimuli. These stimuli are a great way to test memory since they cannot be explicitly/ consciously rehearsed. Further, understanding how meaningless perceptual stimuli are encoded and stored provides a template for all forms of perceptual learning during infancy. Evidence for our ability to detect statistical regularities in meaningless information comes from a study by Guttman and Jules who showed that participants can detect the recurrence of some ‘features’ when presented with a segment of auditory white noise played back to back continuously (cyclic noise, CN)(Guttman, N., and Julesz, 1963). Recent research has shown that humans are remarkably good at remembering meaningless auditory patterns over several weeks without conscious rehearsal (Agus, Thorpe, & Pressnitzer, 2010). They asked participants to discriminate CNs and non-cyclic noise (N), implicitly presenting some CNs several times within a block. Participants improved at detecting cyclicity and retained this knowledge over 2-3 weeks. In this series of experiments, we were interested in exploring the limits, robustness and neural correlates of memory for meaningless sounds.
Experiment : Robustness of memory for implicitly learned Gaussian sounds. In this experiment, we wanted to explore if participants can remember modified versions of a learned Gaussian noise using an experimental setup similar to that used by Agus et al. To test this, participants were presented with looped and scrambled versions (scrambled at 10 or 20 ms bin size) of a learned cyclic noise 4 weeks after the learning session.
Results: Extending the findings of Agus et al (2010), we show that participants can store memories of implicitly learned patterns up to 4 weeks. Participants were significantly better at detecting patterns in intact, looped and scrambled versions of a learned sound compared to new cyclic noises suggesting that neurons might be coding for very small bits of information in the auditory modality.
Plain Noise – 5 sec –
Cyclic Noise – 5 sec –