Neural Substrates of Rule Retrieval & Implementation

1. Neuro-cognitive bases of task control Leipzig, June 5th, 2004 Neural Substrates of Rule Retrieval & Implementation Silvia A. Bunge Department of Psychology & Center for Mind and Brain University of California, Davis http://mindbrain.ucdavis.edu/content/Labs/Bunge/

2. Carter Wendelken, Ph.D. Sarah Donohue Eveline Crone, Ph.D. Acknowledgements

3. How do we suppress interference and override inappropriate responses? • Brain, 2001 • Neuron, 2002 • NeuroImage, 2002 • Neuropsychologia, 2003 (Hazeltine) • J Cog Neurosci, 2002 (Ochsner) • How do we represent goals and rules for behavior? • J Neurophys, 2003 • In prep. • How do basic control processes give rise to higher cognitive functions? • PNAS, 2000 (dual-task performance) • Cerebral Cortex, in press (integration) Cognitive Control Laboratory, UC Davis

4. Unspoken rules for social interaction Explicit, symbolic rules Much of our behavior is guided by rules

5. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving rule retrieval vs. switching • Neural changes underlying the development of rule use?

6. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving rule retrieval vs. switching • Neural changes underlying the development of rule use?

7. PFC lesions: deficit in learning and implementing rules Milner, 1963; Petrides 1985; Passingham, 1993 abstract Wallis, Anderson & Miller, 2001 • PFC neurons maintain rule representations • BUT: spared rule articulation Shallice and Burgess, 1991 Asaad et al., 1998; White & Wise, 1999 • knowledge vs. implementation vs. strategic retrieval: Sylvester & Shimamura, 2002 Gershberg & Shimamura, 1995 PFC and Rule Representation

8. DLPFC : ‘context’ maintenance O’Reilly, Braver & Cohen, 1999 Macdonald et al., 2000: instruction-related activ. of DLPFC in Stroop task • VLPFC : learn & use simple S-R associations Petrides & Milner 1982; Toni 1999; Passingham 2000; Murray 2000 Candidates for Abstract Rule Maintenance What about maintenance of sets of response contingencies?

9. Right Compound2 Left Simple1 Left Simple2 Right Task Rules Sample/Probe Response Left Compound1 Right

10. Rule retrieval Rule maintenance Delay1 Sample Delay2 Probe + + 7 - 15 0.5 2.0 1.5 Trial Structure Cue 1.0 Time (sec) - Regions involved in rule representation: compound > simple - Abstract: not dependent on cue type (shape/verbal) Bunge, Kahn, Wallis, Miller & Wagner, J Neurophys, 2003

11. Posterior middle temporal: rule storage? L anterior VLPFC rule retrieval? Rule Retrieval Compound > Simple Cue period Bunge, Kahn, Wallis, Miller & Wagner, J Neurophys, 2003

12. Rule sensitivity Cue sensitivity 0.4 0.4 Simple Compound1 Shape 0.2 Peak amplitude 0.2 Compound2 Verbal 0 0 Abstract Rule Maintenance parietal post. VLPFC

13. Summary L posterior VLPFC, parietal: rule maintenance • DLPFC: • rule-sensitive • during cue period, • but p > .001 • not rule-sensitive • during delay Posterior Middle temporal: rule storage? anterior VLPFC: rule retrieval? Bunge, Kahn, Wallis, Miller & Wagner, J Neurophys, 2003

14. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving rule retrieval vs. switching • Neural changes underlying the development of rule use?

15. Posterior MTG: Action Knowledge Store? Martin & Chao, 2001 • active when view/name/imagine/retrieve info about • manipulable objects (e.g. tools)

16. 1st UCD study! Questions Posterior middle temporal gyrus • general role in storing action-related • knowledge, including rule meanings? • PFC involvement in retrieval of recently learned vs. well-learned rules?

17. Old Merge (U.S.A.) Trained Wait for counter-traffic (Italy) New Untrained Expressway entrance (Germany) Traffic Signs: study phase • 14 Americans with 4+ yrs driving experience in U.S. • no experience driving abroad • Each sign viewed 1 time for 4 sec • Signs in ‘New-Trained’ category explained 4 times

18. “Whenever a new sign appears on the screen, think about its meaning.” • no response requirements Traffic Signs: fMRI + + … 5 sec variable 5 sec variable

19. Explain meaning Button press: High/Low/Guess • Accuracy • Confidence judgement Traffic Signs: post-test

20. Confidence judgement 0.8 * 0.4 *** 0.8 Proportion of trials 0 0.4 High Low Guess 0 Behavioral Results New-Untrained New-Trained Old Correct identification Proportion of trials N = 14 • Old vs. New-Trained: don’t differ much in terms of • knowledge (only experience) • New-Untrained signs are familiar but not meaningful

21. p <.001 “Passive” viewing of traffic signs All conditions > fixation • Left VLPFC is active during task performance • follow-up analyses: activation doesn’t depend on knowledge or amount of experience

22. Posterior MTG p < .001 Recovering rule meaning: Posterior Middle Temporal Gyrus Retrieve+ > Retrieve- • Bilateral middle temporal gyrus is engaged for meaningful signs, regardless of amount of experience

23. Retrieve+ Old > New-Trained Conjunction p <.0001 p <.005 Recovering rule meaning: Medial Temporal Lobes Retrieve+ > Retrieve- Parahipp. p <.001 • Left parahippocampal cortex: • retrieval of meaningful signs • more active for signs with richer associations

24. R insula (anterior VLPFC) p < .005 Controlled rule retrieval Retrieve+New-trained > Old • Right VLPFC (BA 47) is more active during correct retrieval of rules with weaker than stronger associations

25. Summary • L & R posterior middle temporal: • Retrieve+ >Retrieve- • rule storage • L parahippocampal c.: • Retrieve+ >Retrieve- • Old > New-Trained (Retrieve+) • Automatic rule retrieval • R anterior VLPFC: • New-Trained > Old (Retrieve+) • controlled rule retrieval

26. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving rule retrieval vs. switching • Neural changes underlying the development of rule use?

27. Development of rule use • Developmental improvements in: • representation of rules w contingencies(Zelazo et al., 2003) • rule maintenance over a delay(Diamond, 1991) • flexible rule-switching(Kray et al., 2004) • These functions may mature at different times (Crone et al., in press)

28. rule retrieval: bivalent vs. univalent targets • rule maintenance: 500 vs. 4500 ms cue-target delay • rule switching vs. rule repetition Bivalent targets • 8-year-olds (N = 16) • 13-year-olds (N = 20) • Adults (N = 20) Univalent targets Behavioral study: Development of rule use Crone & Bunge, in prep.

29. Rule retrieval (bivalent vs. univalent targets) Rule switches vs. repetitions Summary of behavioral findings • no group differences in accuracy for rules with univalent targets (see also Zelazo, 2003) • 8 & 13-year-old children less accurate for bivalent rules than adults, and show disproportionally larger RT increases for bivalent > univalent rules • Rule switches less accurate than rule repetitions, but no group diffs • 8-year-olds had proportionally larger switch costs than 13-year-olds or adults when switching to a bivalent (but not univalent) rule. • Different developmental trajectories: • rule switching is mature by 13 • rule retrieval is not mature by 13 Crone & Bunge, in prep.

30. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving rule retrieval vs. switching • Neural changes underlying the development of rule use?

31. . . . Bivalent1 Rest Univalent Rest (20 trials) (20 trials) fMRI study BLOCKED Scan Minimal differences in performance across age groups: good baseline for developmental diff’s MIXED Scans • 90 trials intermixed • 30 trials/rule; ** rule switch vs. repetitions CUE (1 sec) DELAY (.5 sec) TARGET (2.5 sec) • Goal: 60 subjects, ages 8-25 • Current N: • -17 adults (ages 18-25) • - 9 adolescents (ages 13-17) • - 9 children (ages 8-12)

32. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving rule retrieval vs. switching • Neural changes underlying the development of rule use

33. Neural basis of task-switching • fMRI studies variably implicate medial PFC/parietal/lateral PFC • (Dove 2000; Sohn 2000; Brass 2002; Dreher 2002; Bunge 2003; Sylvester 2003) • TMS of medial PFC or parietal cortex impairs switching • (Rushworth 2001, 2002) • lateral PFC: • perseveration on WCST more common w/ damage to medial • than lateral PFC (Stuss 2000) • not reliably observed in fMRI studies (e.g. Braver 2003; Bunge 2003) • active at onset of switch & repeat trials (Dreher 2002; Braver 2002) • rule retrieval (Bunge et al. 2003, in prep; Brass et al., 2003) • involved in rule retrieval rather than switching per se?

34. p < .005 16 adults Rule retrieval vs. switching: Partially overlapping regions? Bivalent > Univalent rules Rule switch > repetition (bivalent rules) Both

35. - L anterior VLPFC (insula; BA 47) - L parietal (BA 7/40), L pre-SMA (BA 6), R pre/motor, basal ganglia p < .005 16 adults A closer look… Bivalent > Univalent rules, excluding Switch trials Rule switch > repetition (bivalent rules)

36. Effect of Rule Type on Switch Costs • Bivalent vs. Univalent targets: • slower, less accurate • greater switch costs

37. Univalent rules N.S. Effect of rule switching is evident for bivalent rules Effect of Rule Type on Switch-Related Activation Switch > Repetition Bivalent rules p < .001 16 adults

38. Tentative summary: adult fMRI data • rule retrieval and rule-switching may be neurally separable • rule retrieval: VLPFC • switching: parietal (7/40), pre-SMA (BA 6), pre/motor, b.ganglia • BUT mechanisms affecting rule retrieval influence • rule-switching, and vice versa: • task-switching is taxed more when switching • to a bivalent rule than to a univalent rule • rule retrieval is taxed more when trials are mixed • than blocked

39. Retrieving and Using Rules for Behavior • Rule retrieval and maintenance • Long-term storage of rules • Maturational timecourse of rule use in children • Brain regions subserving task switching vs. rule retrieval • Neural changes underlying the development of rule use?

40. Rule representation vs. task-switching • Different developmental trajectories, based on differential • maturation of regions involved in rule representation/switching? • (Crone et al., in press; Crone & Bunge, in prep.) Prediction: task switch-related activation (e.g., medial PFC/parietal) matures earlier than rule retrieval-related activation in VLPFC

41. Acknowledgements UC Davis MIT Anthony Wagner Itamar Kahn Earl Miller Jonathan Wallis Eveline Crone Sarah Donohue Carter Wendelken Omri Gillath Michael Souza Jesse Edelstein Additional thanks to… Ron Mangun Charan Ranganath Cameron Carter Barry Giesbrecht Tamara Swaab Russ Poldrack Mike Cohen Craig Brozinsky Funding National Science Foundation (A.W.) New Faculty Research Grant (S.B.)

43. Development of cognitive control

44. Congruent Incongruent Neutral Response selection NO-GO Subjects • 16 children (9 males; ages 8 - 12; M = 10) • 16 adults (9 males; ages 19 - 33; M = 24) Task

45. Adults Kids Behavioral results • Flanker accuracy: ≥ 98% for both groups • Increased interference-related slowing in children Incongruent > Neutral

46. Response selection Incongruent > Neutral

47. Correlations with response selection efficiency Adults Children Magnitude of activation Response selection efficiency better worse ADHD data

48. Different approach to task? “right” • greater interference in children due to slower / less efficient S-R translation (Ridderinkhof et al. 1997) In children, response suppression efficiency: • correlated with fluid verbal ability (WISC-word attack) • NOT correl. w speed of processing or visuospatial skills … related to intermediate step of verbalization?

49. Right VLPFC Selection Inhibition

50. Right VLPFC