Understanding Minds: Disentangling ToM

1. 1 Human Social Interaction Research Proposal Dr. Roger Newport Room B47 Drop-in Times: Tuesdays 12-1 & 3-4 roger.newport@nottingham.ac.uk www.psychology.nottingham.ac.uk/staff/rwn Understanding Minds: Disentangling ToM

2. + your marks

3. Stone and Gerrans (2006) The outputs of lower-level (domain-specific) mechanisms are used for inferences by higher level (domain-general) mechanisms 2 Executive function Joint attention Gaze monitoring Metarepresentation Recursion Emotion recognition Biological motion Face processing EF - keeps social elements in mind & inhibits knowledge of reality M - operates on eye gaze etc. - who saw what R - operates on Ms of mental states - allows reasoning about not just other’s thoughts, but other’s thoughts about other’s thoughts

4. In recent weeks we have looked at domain-specific lower-level mechanisms crucial to human social interaction such as emotion recognition, eye gaze detection, direction of attention detection, biological motion detection and action recognition They are dependent on domain-specific neural circuitry and their normal functioning is an essential precursor to normal ToM performance. But are they sufficient by themselves for sophisticated ToM mechanisms to process social information? Deficits on ToM tasks may result from deficits in low-level social input systems (e.g. joint attention) or in higher-level mechanisms (e.g. executive function). So is it possible to find a pure independent ToM deficit or single ToM mechanism anywhere in the brain? 3

5. 4 Previously: argued about ST & TT and identified different networks apparently associated with each. Big argument as to which is right But not surprising that regions are different because the tasks they are performing are completely different and are measured by different types of tasks. Stop worrying about which is right and think about how they might interact or overlap

6. Integrating ST and TT Keysers and Gazzola, 2007 The simulation camp: Tend to focus on intuitive examples Responding to expressions of e.g disgust Feeling another’s pain Seeing another’s actions Involve shared circuits in e.g. Amygdala/ posterior insula SII PMC &PC } They tend to be measured by by fast responses to observable states of others (through face stimuli, observing grasping etc.) Brain activity correlates with empathy scores and intact structures allow us to know things like whether someone is enjoying a nice cup of tea. 5

7. Integrating ST and TT Keysers and Gazzola, 2007 The theory (of mind) camp: Tend to focus on conscious reflective examples Whether someone has a false belief Details of their inner lives Details of their inner states and introspection Involve non-shared circuits in e.g. Midline frontal structures Temporal regions Anterior Insula } They tend to be measured by by slow responses to thoughts about states and lives of others (through stories, cartoons etc.) Brain activity correlates with thinking about others and our interactions with others. 6

8. 7 Last week: Problems separating ST and TT Most behavioural data can be explained adequately by both This is frustrating Can imaging separate ST from TT?

9. 8 Separating ST from TT using imaging TT ST Predictions about what others will think/do are based, in part, on simulations of our own behaviour Predictions about what others will think/do are based on general rules that govern behaviour - independent of our own actual behaviour Rely on shared brain structures for self and other (not nec. MNs) Do not rely on shared structures Common activation for self v other judgments = ST Separate activation for self v other judgments = TT But it is not enough simply to show similar activation for self vs other: there must also be a theory of mind component

10. Separating ST & TT: Ramnani and Miall, 2004 9 Ss Shown coloured shape Colour = who responds (you, confederate or computer) Shape = which key response Task: to monitor errors of other and computer Therefore can see brain regions active for preparation of self-produced responses and anticipation of other-produced responses Activation for both self and other in premotor areas (so using motor areas to monitor anticipated other responses as well as self responses), but areas did not overlap suggesting TT, rather than ST But, where’s the ToM component? Task does not require understanding beliefs, desires or intentions of other

11. Grezes et al., 2004 False belief from action observation 10 Task: Watch pre-recorded videos of self or other picking up a box having been told, sometimes incorrectly, the weight of the box. Found bilateral overlapping premotor activation for viewing self and other (although activation faster for self) Does this = ST? Not a true self vs. other judgment. Because Ss are watching videos of themselves. This task involves making a 3rd person judgment about themselves rather than a 1st person judgment No evidence for ToM component (simply action observation) + comparing true vs. false belief trials would subtract out belief reasoning, leaving just the observation of a discrepancy between the expected and observed outcomes of the movement Also don’t know if target had a false (or any) belief

12. Separating ST & TT using ToM stories: Vogeley, 2001 11 Task: Ss listened to stories while in scanner that involved either thinking about others’ thoughts or their own thoughts e.g. A shop has been robbed, the policeman comes etc. (ToM) vs. Your shop has been robbed and you cannot stop them (no mentalising) What does the policeman do? vs. what do you do? All compared to physical causality stories Found some overlapping areas active for ToM regardless of self or other (ACC, left temporal pole) = ST Area more active for thoughts about self (right PFC) = distinct TT Has ToM component and is compared to non-ToM for subtraction But not true self v other task. Making 3rd person judgments about fictional self. Compares 3rd person self (fictional situations) to 3rd person other rather than 1st person self to 3rd person other Without 1st person component you cannot test ST as ST requires 1st person simulation to generate 3rd person decisions.

13. 12 So, interim summary Experiments designed to separate ST from TT must have: a) ToM component (that is not subtracted out) b) 1st person self vs 3rd person other comparison + you must be sure when/if 1st person beliefs and desires are being formulated in your Ss. Good luck with that. According to Apperly(2007) this may prove impossible Alternative, more productive routes: May be better off trying to separate different components of theory of mind OR If we can’t measure separable ST & TT, can we measure integration of ST & TT?

14. Integrating ST and TT Keysers and Gazzola, 2007 reflective Intuitive/pre-reflective Shared Circuits (e.g. MNs) Midline structures etc Nice cup of tea Nice socks for Christmas Grasping experiments Thinking experiments 13

15. Some of the brain regions we are likely to come across: 9 dlpfc 10 pfc 32/24 ACC 11 ofc 9dpfc 22/39 STS/TPJ 10 pfc 11 ofc 14

16. When I think about me and simulate you: Medial rostral prefrontal cortex and self-referential processes Benoit et al. 2010 NeuroImage Self-referential processing (us referring things to ourselves) We like to think of ourselves as constant and coherent entities We can attribute personality traits and personal memories to ourselves 15

17. mrPFC has been implicated in tasks thought to require self-referential processing e.g. judging whether personality trait words are descriptive of oneself (more active if they are) mrPFC also more strongly activated during such judgments about oneself than about either close friends or famous others But… this region is not exclusively recruited while thinking about oneself (self-referencing) Also get activation during trait attributions about either oneself or other people 16

18. Simulation accounts suggest that one can make inferences about others by referring to knowledge about oneself (e.g. extrapolate own thoughts and feelings to what another person might think and feel). only valid if the self is a good model for the other – have to assume that the other person is reasonably similar to oneself. If it is then mrPFC might also be involved in trait judgments about others (see Mitchell et al., 2005 - more similar activation of this region during mental state inferences about oneself and an unknown person, if this unknown person was thought of as more politically like-minded). 17

19. tested whether mrPFC is also more strongly engaged during trait judgments about personally known others in cases of greater perceived similarity. recruiting self-referential processes for similar others may impact on performance on a subsequent episodic source memory task: if trait attributions about the self and a similar friend are subserved by common processes, these shared encoding operations might yield overlapping memory representations for both judgment types remembering the reference of the judgments was expected to be more difficult in case of greater similarity. 18

20. Task 1: Rate (1-4) how much these apply to you (self cond) or best friend (other cond) Task 2: ‘Yes’ to old/same and ‘no’ to old/different & new words for relative condition in Task 1 Control task = syllable counting 19

21. 20 Blue = self Red = other mrPFC was activated during appraisal of one's personality traits as well as during subsequent retrieval of those judgments. Crucially, the degree to which this area was also activated while processing others correlated with the degree of perceived similarity. Thus, the data support simulation accounts which posit that self-referential processes are also employed for thinking about another person, if one assumes oneself to be a good model of this person. Moreover, recruiting similar processes for appraising oneself and others was associated with impaired source memory for these judgments, which suggests an impact of extrapolating self-referential processes on subsequent behavior.

22. Separation of self and other in 3rd person thinking 21 Simulation is the same as self-reflection You simulate what you would do in a certain situation We use this strategy when the 3rd person we are thinking about is similar to ourselves Keysers and Gazzola, 2007

23. Separation of self and other in 3rd person thinking 22 Mitchell et al., 2005 & 2006 fMRI task: look at photos and make social (how happy to have photo taken) vssymmetricity judgment. Behavioural exp: rate how similar people in photos are to self Making judgments (i.e. thinking) about similar people (using simulation?) activates vmPFC Thinking about dissimilar people activated dmPFC BUT, introspection or self-reflection is theoretically a pre-requisite for both ST and TT - it cannot differentiate them And simulation doesn’t necessarily rest on the similarity of ourselves and others. We can run a simulation based on the fact that our target is different.

24. 23 For successful HSI we must be able to recognize that the behaviour of other people is ruled by the same sort of mental states—beliefs, desires, intentions—that guide our own behaviour. Mitchell’s idea was to use fMRI to see how perceivers make mental state inferences when the self-other overlap can be assumed (when the other is similar to oneself) and when it cannot (when the other is dissimilar from oneself).

25. 24 dmpfc activation is often associated with mentalising tasks, but not self-referencing, whereas vmpfc is more commonly associated with self-referencing But vmpfc is occasionally recruited in mentalising tasks – especially when thinking about similar others • Could it be that vmpfc activation in ToM tasks reflects simulation of self attributes when thinking about others? – that is, when the inferences about the mental states of others are informed by the perceiver’s knowledge about their own thoughts and feelings

26. Prior to scanning, participants were introduced to two target individuals (faces) Each face had a short description of a similar or dissimilar person (liberal of conservative views) During scanning, participants judged how likely targets were to agree with each of 66 opinion questions presented below one of three targets: the liberal, the conservative and a silhouette with the word “me”. Questions like “to look forward to going home for Thanksgiving?”; “to enjoy having a roommate from a different country?”; “to drive a small car entirely for environmental reasons?”; “to think that European films are generally better than the ones made in Hollywood?”; “to believe that cultural diversity should be an important national issue?”; etc.). Participants then completed a version of the Implicit Association Test 25

27. 26 Rate Ppts on political IAT

28. 27 Greater activation in vmPFC for thinking about those more similar Greater activation in dmPFC for thinking about those more dissimilar No difference for self “We observed a double dissociation … mentalizing about a similar other engaged a region of ventral mPFC linked to self-referential thought, … mentalizing about a dissimilar other engaged a more dorsal region of mPFC. … overlap between judgments of self and similar others suggests the plausibility of “simulation” accounts … which posit that perceivers can use knowledge about themselves to infer the mental states of others.” Mitchell

29. Let’s have a break

30. Forget about ST vs TT Dissociating different aspects of ToM Samson et al., 2007 ToM requires inhibition of self-perspectives and the engagement of other-perspectives Is one simply more difficult than the other. Task: self vs other perspective taking 28

31. Samson et al., 2007: High vs Low self-knowledge inhibition F M 29 Video 1: F sees M put ball in box F leaves We see M move ball (but not where to/from) F comes back F indicates box she believes ball is A B Video 2: F sees M put ball in box F leaves We see M move ball (we do see where to) F comes back F indicates box she believes ball is Q. Where is the ball? We don’t know where the ball is so answer requires knowing that F has a false belief Q. Where does the woman think the ball is? We know and we know the woman has a false belief. The answer requires knowing that F has a false belief AND inhibiting our own knowledge

32. Dissociating different aspects of ToMSamson et al., 2007 ToM requires inhibition of self-perspectives and engaging other-perspectives Task: self vs other perspective taking 2 patients: WBA - RH stroke including right PFC - scored poorly on false belief tasks requiring self-knowledge inhibition PF - LH stroke involving left TPJ (inc left anggyrus and left STG) WBA - passed low inhibition / failed high inhibition PF - failed low inhibition / passed high inhibition (but directed by verbal nature of task to pay attention to Fs false belief) WBA (rPFC) has self-perspective bias - impairment in self inhibition - a domain-general impairment PF (TPJ) - does not know to automatically apply false belief knowledge (but can if directed) - domain-specific impairment 30

33. Sommer et al., 2007. ToMfMRI using the correct controls False beliefs are necessary for ToM Previous studies have implicated either (or both) the TPJ and mPFC in (false) belief reasoning. While developmental studies have contrasted faslev true belief reasoning in children, adult studies of false belief reasoning have not (and have therefore used incorrect control conditions - e.g. physical stories (no belief required), cartoons (may involve mentalising) etc. Therefore, to address this, Sommer et al. applied false v true belief task in adults using fMRI 31

34. 32 Story Proposed answers Attribution of intention condition Physical causality with characters Physical causality with objects

35. The True Belief Task 33 Control pictures (same visual complexity as belief pics) Belief pictures Behavioural only pictures

36. The False Belief Task 34 Analysis: Control v true; control v false; false v true; true v false

37. 35 Results False beliefs = dACC (/mPFC) AND rTPJ (probably doing different aspects of task): TPJ = perspective taking; ACC = situation monitoring Beliefs in general not associated with mPFC or TPJ.

38. Verbal v non-verbal ToM in children and adults Kobayashi et al., 2007 Theory of Mind tasks usually highlight mPFC and/or TPJ. Some experiments use verbal tasks, some non-verbal tasks. Some experiments use children and some adults. In order to determine the modality-specific (verbal v non-verbal) development of ToM we must use both tasks in both age groups. 36

39. Verbal v non-verbal ToM in children and adults Kobayashi et al., 2007 Verbal Non verbal 37

40. Verbal v non-verbal ToM in children and adults Kobayashi et al., 2007 Also used non-ToM conditions e.g. ToM Non-ToM 38

41. Verbal v non-verbal ToM in children and adults Kobayashi et al., 2007 Theory of Mind tasks usually highlight mPFC and/or TPJ. Some experiments use verbal tasks, some non-verbal tasks. Some experiments use children and some adults. In order to determine the modality-specific (verbal v non-verbal) development of ToM we must use both tasks in both age groups. ToM tasks caused more activity in rPFC, rIPL, bilat TPJ Greater STG for story vs cartoon task. Children ToM greater than adult ToM in rPFC and rSTG Therefore cognitive resources used by adults and children are different and develop dynamically. 39

42. Multi-level comparison of empathy in schizophrenia: An fMRI study of a cartoon task Lee et al., 2010 Psychiatry Research: Neuroimaging 40 Schz are impaired at social functioning Empathy is an essential part of social functioning Empathy requires/ can be broken down into several components: Cognitive empathy – the ability to see/understand the world from someone else’s point of view Affective empathy – the ability to share emotional experiences Origin regulation – keeping track of self and other’s feelings/thoughts Cognitive empathy sometimes referred to a ToM.

43. 41 In this study they tried to identify the neural structures involved in HCS and Schz for Cognitive empathy (understanding others’ thoughts) Emotional empathy (understanding others’ feelings) Inhibitory empathy (empathise by suppressing others’ intentions/feelings) Physical causality control We are really only interested in cognitive empathy

44. 42

45. 43 CE-PC Main difference is lack of activation in temporal poles (reduced volume in temporal region also previously found in Schz) Temporal poles associated with retrieval of personal situations/events and with recalling familiar faces So perhaps Schz impaired at memory for past social information that could be used to have conceptual understanding of social situations (inc. others’ beliefs, desires, feelings etc.).

46. Theory of mind deficits in patients with acquired brain injury: a quantitative review Martin-Rodrıguez & Leon-Carrion. 2010. Neuropsychologia 44 First order theory of mind tasks (FOTOM) Like the Sally-Ann task: A thinks X. Is this a true or false belief? Passed by age 4 Second order theory of mind tasks (SOTOM) Recursion: A thinks B thinks X. More complex: One character can have beliefs about another character Passed by age 6 Increasing difficulty Indirect Speech: understanding irony, sarcasm, metaphors, or jokes. Involve SOTOM involving language – inferring the hidden intention behind statements Passed by 7-9 Social faux-pas: Knowing someone has said the wrong thing without realising they have done so. Includes cognitive and emotional ToM aspects. Passed by 9

47. 45 Meta analysis searched the MEDLINE and PsychINFO databases using the keywords: “theory of mind” [AND] “brain injury”, “theory of mind” [AND] “brain damage”, and “theory of mind” [AND] “head injury”, limited to ABI patients who had suffered brain injury in adulthood. only studies that compared patients’ ToM performance with healthy controls were considered. Only studies that used FOTOM, SOTOM, understanding IS and detecting faux pas tasks to evaluate ToM abilities were included

48. 46 Results moderate to severe impairment in ToM reasoning among patients with ABI: highest effect size for understanding IS, followed by faux pas, SOTOM, and finally FOTOM; but with considerable overlap. Different to difficulty hierarchy Severity of impairment influenced by proportion of patients with frontal and right H damage (after all that - not all that informative) Fits well with recent rTMS experiment showing decrement in performance (RT & accuracy) in healthy controls on False belief and faux pas tasks respectively. (1 hzrTMS over left/right dorsolateral prefrontal cortex and rTPJ)

49. Live face-to-face interaction during fMRI: A new tool for social cognitive neuroscience Elizabeth Redcay et al. 2010. NeuroImage 47 functional MRI data were collected while participants interacted with a human experimenter face-to-face via live video feed as they engaged in simple cooperative games. In Experiment 1, participants engaged in a live interaction with the experimenter (“Live”) or watched a video of the same interaction (“Recorded”). During the “Live” interaction, as compared to the Recorded conditions, greater activation was seen in brain regions involved in social cognition and reward, including the right temporoparietal junction (rTPJ), anterior cingulate cortex (ACC), right superior temporal sulcus (rSTS), ventral striatum, and amygdala This method of allowing online, contingent social interactions in the scanner could open up new avenues of research in social cognitive neuroscience, both in typical and atypical populations.

50. Summary Trying to separate or identify ST vs TT processes is difficult (but theoretically possible) Think about 1st vs 3rd person beliefs and perspectives Think about overlap/separation of self and other Consider carefully whether you have a ToM component Separating ToM components e.g.: Verbal v non-verbal Domain-general v domain-specific Perspective taking Beliefs Plus many many more 47