2. The Cognitive Neuroscience Of True and False Memories Daniel L. Schacter Harvard University

3. “Against All Enemies” Mr Clarke: The Secret Service asked for fighter escorts to protect Air Force One. Mr. Miller: He himself asked Condoleeza Rice whether to call up fighter support and she said to go ahead. Mr. Clarke: Mr. Miller urged Secretary Rumsfeld to take a helicopter out of the Pentagon. Mr. Miller: He never talked to Mr. Rumsfeld that day.

9. Cognitive Neuroscience of Memory Distortion (Schacter & Slotnick, Neuron, 2004) *What brain regions are involved in monitoring or reducing memory errors? * What type of brain activity is associated with the generation of inaccurate memories? *Can brain activity distinguish true vs. false memories? The sensory reactivation hypothesis *What are the neural origins of memory errors?

11. Norman & Schacter, 1997

15. Sensory Reactivation Hypothesis *True memories are characterized by greater access to sensory/perceptual details than are false memories; reflects reactivation of sensory/perceptual encoding for previously experienced events. *Behavioral evidence: Studies of post-event misinformation (Schooler et al., 1986) and false recognition (Mather et al., 1997; Norman & Schacter, 1997): access to sensory/perceptual details greater for true than false memories.

16. Sensory Reactivation Hypothesis: Neuroimaging Evidence *True>false increases in DRM paradigm: Subjects intially hear lists of semantically related words, tested visually with old and new items True>false increases at retrieval in auditory cortex suggestive of auditory reactivation (Schacter et al., 1996). *However, later research suggests that true>false activation may depend on exact format of recognition test (blocked v. intermixed; Johnson et al., 1997; Schacter et al., 1997).

17. FMRI of True and False Recognition: Basic Procedure (Cabeza et al., 2001) Participants study lists of semantically related words (half of the lists are semantic associates from Roediger & McDermott, half are categorized lists) Each list presented alternatively on videotape by either a male or female source Participants instructed to try to remember words as well as which speaker presented them Scanned during old/new recognition test

19. True vs. False Recognition of Shapes (Slotnick & Schacter, Nat. Neurosci, 2004) * Previous results suggest greater sensory reactivation during true recognition compared with false recognition. * We tested this idea by assessing event-related activity in the visual cortical processing stream during true and false recognition of visual shapes.

20. Study Phase  For each of 3 runs at study, 144 shapes were presented (16 sets of 9 exemplars)  Each set alternated in spatial position to the right or left of fixation  Pres. Time = 2.5 sec  Instructions: remember each shape and side of the screen Nonstudied Prototype Exemplar

21. Test Phase  For each of 3 test lists, 96 shapes:16 sets of 2 studied exemplars, 2 nonstudied/related shapes (1 prototype and 1 exemplar), and 2 nonstudied/ unrelated shapes  Presented at center for 2.5 sec  Participants made recognition decision: “old-left”, “old-right”, or “new”

22. Behavioral Results OldRelated New P(“Old”) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 * * * p < 0.001 *

23. True and False Recognition Ventral View Early and late visual regions (BA17, BA18, BA19, BA37) Time (sec) % Signal change 0 0.1 -0.1 0 4 8 1216 Left fusiform gyrus (BA37) Old-hits Related-FAs New-CRs Old-hits > New-CRs (True recognition) + Related-FAs > New-CRs (False recognition)

24. True recognition > False recognition Old-hits > Related-false alarms Related-false alarms > Old-hits Ventral View LH Early visual regions (BA17, BA18) Time (sec) % Signal change 0 0.1 0.2 0.3 -0.1 -0.2 -0.3 0 4 8 1216 Left fusiform gyrus (BA18) Old-hits Related-false alarms * Visual recognition- related activity previously reported in late visual areas (BA19, BA37) X

25. Sensory Reactivation Hypothesis: Nature of Visual Activity? *Results show greater activation for true than false memories in early visual areas, which process primitive visual information. *Does this reflect conscious memory for previously studied shapes? Perhaps recollection of some feature of a shape - a line segment or color? *NO!

26. Sensory Reactivation Hypothesis: Nature of Visual Activity? *Early visual activation is the same for previously studied shapes both when people call the shape “old” (indicating memory) and when they call it “new” (indicating forgetting). *This finding suggests that a nonconscious form of memory distinguishes between true and false recognition. *In contrast, later visual areas that activate similarly for true and false recognition do reflect conscious memory.

27. Cognitive Neuroscience of Memory Distortion (Schacter & Slotnick, Neuron, 2004) *What brain regions are involved in monitoring or reducing memory errors? * What type of brain activity is associated with the generation of inaccurate memories? *Can brain activity distinguish true vs. false memories? The sensory reactivation hypothesis *What are the neural origins of memory errors?

28. Cognitive Neuroscience of Memory Distortion: Encoding Origins *What are the neural origins of memory distortion? *Can we tell by examining brain activity at the time of encoding whether an event will be remembered accurately or inaccurately? *Several recent studies have begun to explore this question using subsequent memory designs, where scanning brain activity at the time of encoding allows us to predict whether an item will later be remembered or forgotten.

29. Wagner et al., Science, 1998

31. Specific & Nonspecific Recognition Accurate memory is often highly specific, whereas distorted memories rely more on general “gist” of what happened: – Specific recognition = exact or verbatim memory for previously studied item – Nonspecific recognition = general memory for gist of previously studied item (Brainerd & Reyna, 1995; Schacter et al., 1998) e.g., false recognition: calling related new item “old”

32. Specific vs. Nonspecific Recognition: Rationale We sought to compare: * the neural origins of specific and nonspecific recognition at encoding using a subsequent memory design *Key hypothesis: activity during encoding in the right fusiform region, implicated in processing specific visual features of objects, will be preferentially associated with subsequent specific recognition (Garoff, Slotnick, & Schacter, Neuropsychologia, 2005)

33. Method Encoding During fMRI scanning Participants view 360 nameable objects Task – Is object larger or smaller than a 13-inch square box?

34. Method Recognition Next day, outside of scanner Participants view 450 objects (180 same, 180 similar, 90 new) Task – Is object same, similar, or new

35. fMRI Results – Distinct Neural Regions Z = -20 right fusiform cortex samesimilarnew “same” specific recog false recog “similar” partial recog “new” * Supports hypothesis that right fusiform gyrus is associated with specific feature encoding *Right fusiform gyrus is ONLY area to show activity in this contrast Specific recognition > Nonspecific recognition

36. Reality Monitoring and the Brain *Reality monitoring refers to our ability to distinguish between memory and imagination: remembering perceived events versus things we only imagined (Johnson & Raye, 1981) *Studies using fMRI to measure brain activity during encoding provide evidence that we can predict, to some extent, when reality monitoring will succeed and when it will fail. *We examined reality monitoring for emotional and nonemotional information.

37. frog casket canoe snake Adapted Adapted from Gonsalves and Paller (2000) Paller, 2000 Evoking Reality-Monitoring Errors Was the corresponding picture presented at study? frog fire canoe road snake casket

38. Kensinger & Schacter (2005), Neuropsychologia  Role of encoding processes?  Activity in regions that process emotional information during encoding increased the likelihood of subsequent correct memory attributions for emotional but not nonemotional items - Amygdala & orbitofrontal cortex  Interactions between amygdala and hippocampus  Important for enhancing correct memory attributions for emotional items Memories for emotional events were more accurate than memories for neutral events: better discrimination between seen and imagined objects

39. FMRI of True and False Memory: Implications for Witness Testimony? *Is fMRI a “truth machine” that we can use to determine the accuracy of witness memory? *Not now (or for forseeable future): -Results apply to groups, not individuals. -As experimenters, we know what actually happened; in the real world, we don’t know. -However, our results show that in principle imaging can distinguish true and false memories.