Parallel Contributions of Distinct Human Memory Systems During Probabalistic Learning. Dickerson, Li, Delgado. Neuroimage 2011.

  1. Some theories hold that there is competitive behavior (such as with striatum and hippocampus) <I guess by which they mean one supresses the other altogether, given their definition of parallel>, other theories hold that they are more cooperative, here they consider that they may be parallel, which simply means they both operate based on the same data, but in the end the way the information they processed is used may be cooperative or competitive
  2. Here they test their interaction in an RL task
  3. There are a number of connections between the systems examined here, and different topology is suggestive of differing functions of each component
  4. “Specifically we hypothesized that both the striatum and the hippocampus would be engaged during learning, but rather than showing signs of competition (negative correlations) we predicted observing signs of parallel activity (voxels within both ROIs involved in probability learning irrespective of learning type).”
  5. In the RL task, a number of symbols were presented that either corresponded to a high or low value <The description is weird as they were told if they symbol corresponded to a number higher or lower than 5, although at this point the only information is which category each symbol belongs to.  I imagine there will be an extension later that will make this setup purposeful>
  6. In another experiment the information presented (high/low) was stochastically dependent on the value of the symbol <not described so clearly>
  7. There is a second part of the study where the subjects are told whether or not their response was recorded <weird>
    1. “In the observation version, participants receive a message indicating whether or not their response has been recorded.  It is hypothesized that the feedback version is akin to trial and error learning, considered to be more similar to nondeclarative types of learning, as the participant must learn which cue is associated with what value via guessing (initially) and subsequently receiving feedback”
    2. <This is not trial and error learning.  Trial and error learning would involve a response telling the subject whether the response was correct or not.  I don’t understand the relevance of saying whether a response was recorded in terms of trial and error learning>
  8. <skipping sections on FMRI setup and analysis, onto results>
  9. Trials were easy (85/15 noise) or hard (65/35 noise).  Not surprisingly, significant relationship between difficulty and rate of correct responses
  10. <either I didn’t do the paper justice by reading it correctly, or I don’t understand the basic difference between “obsevered” and “unobserved” episodes, but I’m finding it hard to care because at the moment the whole setup doesn’t make sense.>
  11. <Doesn’t look like the experiement is expanded in a way to get rid of my criticism in #5, too many problems – skipping along to discussion.>
  12. Ultimately, they conclude that the two systems work in parallel, <but again, I’m not convinced yet the experiment tests what they say they are testing.>

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: