Presentation is loading. Please wait.

Presentation is loading. Please wait.

KUHL, B.A., JOHNSON, M.K., AND CHUN, M.M. (2013). DISSOCIABLE NEURAL MECHANISMS FOR GOAL-DIRECTED VERSUS INCIDENTAL MEMORY REACTIVATION. THE JOURNAL OF.

Similar presentations


Presentation on theme: "KUHL, B.A., JOHNSON, M.K., AND CHUN, M.M. (2013). DISSOCIABLE NEURAL MECHANISMS FOR GOAL-DIRECTED VERSUS INCIDENTAL MEMORY REACTIVATION. THE JOURNAL OF."— Presentation transcript:

1 KUHL, B.A., JOHNSON, M.K., AND CHUN, M.M. (2013). DISSOCIABLE NEURAL MECHANISMS FOR GOAL-DIRECTED VERSUS INCIDENTAL MEMORY REACTIVATION. THE JOURNAL OF NEUROSCIENCE. 33(41), 16099-16109.

2 Prior research says… The hippocampus supports binding features into an integrated memory (Squire, 1992). The frontoparietal cortex is believed to bias memory processing toward goal-relevant representations (Miller, 2000). It has been suggest that control regions do not actively represent the contents of memory (Buckner and Wheeler, 2001)

3 Materials and Task 3 phases: Study Test Post-test Participants studied words (e.g., candle) that were paired with well know people (e.g., Elton John) or well known locations (Big Ben). Participants were cued to recall the corresponding image’s category or location. No test served as the baseline. Surprise post-test required participant to recall the category and location. 12 blocks 12 study trials, 8 test trials

4 Study Encode both category and location. Words and images were presented for 4 s followed by an 8 s mental arithmetic task (baseline). 1 s fixation 1 s 2 digit number 1 s plus sign 1 s 2 digit number 2 s number followed by a question mark 50% correct, 50% incorrect Baseline task prevented participants form rehearsing study items between trials.

5 Test Participants were asked to indicate the category or location of the image that was paired with the word presented on the test screen. F = face S = scene L = left R = right DK = don’t know Test trials lasted 4 s and were followed by an 8 s baseline task.

6 Surprise Post-Test! Category and location 2/3 words were tested during the initial test 1/3 were not Definitely face/left Probably face/left Probably scene/right Definitely scene/right Self-paced

7 3 cubic mm voxels smoothed with a Gaussian Kernel 3 broad ROIs: PFC, LPC, VMTL Subregions in each ROI: PFC: inferior, middle, superior, and medial frontal gyri (including ACC), and orbitofrontal cortex LPC: temporoparietal junction, supramarginal gyrus, angular gyrus, and superior parietal lobule, VMTL: fusiform gyri, parahippocampal gyri, and hippocampus

8 Pattern Classifier Broad ROIs: 500 voxels from each region most sensitive to category differences (face vs. scene). Subregion ROIs: 50 voxels most sensitive to category. Classifier trained to discriminate category on all study trials. Classifier tested on all test trials. Classifier performance was assessed by: Accuracy: was the test trial able to accurately classify the studied picture. Used for general assessment Output: probability from 0-1 that the classifier assigned the correct category for the trial. Used when considering relationships between response speed and memory Used when assessing region-to-region correlations.

9 Behavioral Results Memory during the test was successful although memory was better for categories than for locations. Category memory at post-test was better in both the category test condition and the location test condition than the no test condition (baseline). The same was true for location memory. But testing had a greater benefit for the tested feature than the incidental feature.

10 Classifier Accuracy Classifier performance for category: above chance on category trials (goal-directed) and for location trials (incidental) Significant goal-directed modulation in the LPC. Category trial: p <.0001 Location trial: p <.005 LPC category vs. location: p <.005

11 Target reactivation: 4-6 s: more robust reactivation for VMTL than LPC. 10-12 s: more robust reactivation for LPC than VMTL. Incidental reactivation: No time x region interaction. VMTL activity is more transient while LPC is more sustained. PFC was not significantly different than VMTL for target or incidental.

12 LPC Reactivation most robust in ANG and SPL. More goal-relevant modulation compared to goal-irrelevant in the SMG and SPL.

13 VMTL No significant differences in reactivation for goal-directed compared to incidental for all 3 subregions (FG, PHG, HIPP)

14 PFC Although the broad PFC did not find goal-directed modulation, averaging across PFC subregions did reveal significant goal modulation. Goal modulation was also significant in PFC compared to the VMTL when comparing data from subregions. OFC: significantly more goal modulation than irrelevant.

15 Decoding the goal- relevant feature Retrieval goals influenced the pattern activity in the PFC and LPC, but NOT the VMTL. Peak at 8-12 s suggests goal classification was based on retrieved information and not just a tonic goal state.

16 Cross-region correlations Averaging across category and location trials, correlations were significant for all pairings. The correlation between PFC- LPC was significantly stronger than other pairings however.

17 Cross-region correlations PFC and LPC subregions: Robust correlation between SFG and ANG.

18 Cross-region correlations PFC-LPC correlation stable across time. LPC-VMTL and PFC-VMTL correlations peak when VMTL reactivation peaks. PFC and LPC “coupled” with VMTL when VMTL reactivation was high. PFC and LPC content became similar to each other and different from VMTL when VMTL reactivation decreased.

19 Response time and reactivation Faster behavioral responses when VMTL category activity peaked earlier. Faster behavioral responses when PFC category activity peaked earlier or when there was greater overall activation (but only for category trials). PFC reactivation relates to response time when retrieval is goal-relevant. LPC is unrelated to response time.

20 Reactivation and subsequent memory PFC and LPC: category reactivation for category trials (goal-relevant) predicted category memory on the surprise post-test. VMTL: category reactivation for location trials (goal-irrelevant) predicted subsequent category memory. Only data from correct responses in test trials were used

21 Summary PFC and LPC: Greater category reactivation when category information was goal-relevant. Target reactivation in initial test trials predicted memory for the target feature. VMTL: Incidental reactivation was as strong as goal-relevant reactivation. Incidental reactivation predicted memory for incidental features.


Download ppt "KUHL, B.A., JOHNSON, M.K., AND CHUN, M.M. (2013). DISSOCIABLE NEURAL MECHANISMS FOR GOAL-DIRECTED VERSUS INCIDENTAL MEMORY REACTIVATION. THE JOURNAL OF."

Similar presentations


Ads by Google