How do we know what someone is feeling or intending, especially without the benefit of verbal communication?  How do we know that another driver is angry, that our child is tired, that our dinner companion, mouth full, wants us to pass the salad dressing?  Why is it easy to know these things on some occasions-- or even impossible to ignore them-- but difficult at other times?  How do we know that "that thing is alive," and how do our perceptions of an object change when we see it as a living thing?  

I study the neural and psychological mechanisms underlying how we perceive and interpret the minds of others.  My research employs neuropsychological methods (studies of brain-damaged patients) as well as behavioral testing and functional neuroimaging of healthy individuals.  Much of my work focuses on the perception of mental state information based on non-verbal cues, including facial expressions, body postures, and body movements.  From a neuroanatomical perspective, I am particularly interested in the social functions of the ventromedial frontal lobe and the amygdala.

An ongoing interest of mine involves the relationships among the various social cognitive processes that we call empathy, emotion recognition, animacy perception, anthropomorphizing, and agency attribution. Is it possible to see an object as animate, but emotionless?  Or emotional, but inanimate?  Are the neural substrates that enable us to attribute emotions to others the same neural systems involved in the attribution of goals?   Does thinking about someone's goals change the way we think about her emotions?  Answering these questions, in addition to satisfying our collective curiosity, may help us understand mental disorders such as autism.  Furthermore, understanding the components of mind attribution can contribute to the development of training programs for professions in which empathy control is critical, such as in medical practice.

(Please note: Electronic versions of some papers are provided to allow timely dissemination of academic work for individual, noncommercial use.  Copyrights reside with the respective copyright holders, as specified within each document.  These files may not be reposted without permission.)

Current research questions include:

What is the role of ventromedial frontal lobe cortices in storing social, as compared to affective, knowledge? (With Mahzarin Banaji and Lesley Fellows)

This project uses a technique for examining implicit associations between concepts (the Implicit Association Test, or IAT; for more information see: www. projectimplicit.net).  Prior research (Milne & Grafman, 2000) purportedly showed that patients with lesions to ventromedial prefrontal cortex do not show evidence of social stereotypes (specifically, linking the category “male” more with the attribute “strong”, and the category “female” more with the attribute “weak”) .  However, recent work utilizing more careful testing and scoring methods (Wheeler et al., in prep.), while replicating this initial finding, also finds that similar patients do not show affective associations (linking the category “insect” more with the attribute “unpleasant”, and the category “flower” more with the attribute “pleasant”); this implies that VMF cortices are not critical specifically for social associations, but rather for affective ones, whether or not they are social.  To test this hypothesis, we are testing patients with ventromedial prefrontal damage, and comparing them to patients with dorsal and/or lateral prefrontal damage, on a series of IATs to examine associations which are social+affective, social but not (as) affective, affective but not social, and neither affective nor social.

Is information about intention stored with affective knowledge to inform judgments about people?  (With Kristina Olson, Elizabeth Kensinger, Mieke Verfaellie, and Mahzarin Banaji)

This project developed from a finding by Olson and colleagues showing that both adults and children prefer targets who were lucky to those who were unlucky, in the absence of any mitigating information.  We are exploring one potential mechanism for this preference, the Affective Tagging Hypothesis, both in memory studies of neurologically healthy young adults and in studies of amnesic subjects.  In these studies, we examine “liking” ratings given to faces previously paired with short statements describing a good action or bad action (intentional) or a lucky event or unlucky event (random).  Neurologically healthy young adults give higher ratings to faces which had been paired with lucky events, even when they don’t remember what type of event was paired with the face  (Olson, Heberlein, et al., in preparation).  In the next few weeks, we plan to begin testing amnesic subjects—who have been shown to develop preferences for “good” over “bad” people based on prior experience (Tranel & Damasio, 1993) or on vignettes describing actions (Johnson, Kim & Risse, 1985)-- to examine whether a preference for lucky people is similarly retained.

What factors contribute to empathy control, especially as it relates to medical contexts?  (With Dan Wegner)

This project arose out of an interest in top-down vs. bottom-up contributions to empathy, and developed through discussions of medical empathy.  Medical schools and professional medical organizations have been focusing on a perceived lack of physician empathy, and attempting to formulate ways to teach empathy.  We aimed to bring psychological and neuroscientific models of empathy control to this issue.  In an open-ended questionnaire, we probed physicians’ intuitions about their own empathy: When did they wish to be less empathic (i.e. when excessive identification with their patients’ pain or distress impaired their own functioning)?  When did they wish to be more empathic (i.e. in what kinds of situations did they have trouble mustering appropriate levels of concern for their patients’ emotional or experiential state)?  What did they think hampered their ability to up-or down-regulate their empathy in either of these types of situations, and what sorts of strategies worked to overcome such barriers?  Several interesting trends emerged even from the first few responses we received, which link suggestively both with social psychological models of empathy and with neuroimaging studies of pain empathy.  We chose to focus on two of these:

1) focus on goals was frequently suggested as an effective strategy for overcoming distress at a patient’s pain or poor prognosis.  This is interesting in light of recent findings (Gray, Gray & Wegner, 2007) suggesting that attribution of agency or goals and attribution of experience or feelings are distinct components of how we think of other minds.  We are developing experiments addressing the differential effects of focus on goals vs. on experience in empathy for others, as measured via several variables.

2) one physician commented that though she felt faint when viewing very traumatic injuries, she had no such feelings when she knew the patient with the injury was unconscious—which raises interesting questions about interactions between top-down knowledge (patient is unable to feel anything) and what would appear to be an automatic response (wooziness in response to trauma).  To explore this potential interaction, we are developing experiments to address the role of knowledge about consciousness state in empathy for experience, as measured via behavioral (memory, self-rated distress) and autonomic variables.

Papers, published or in press:

Heberlein AS, Adolphs R. (2004)  Perception in the absence of social attribution: selective impairment in anthropomorphizing following bilateral amygala damage.  Proceedings of the National Academy of Sciences 101(19):7487-7491.
(The first three papers listed here were part of my dissertation, under the supervision of Ralph Adolphs, then at the Univ. of Iowa but now at Caltech). We tested a rare patient with bilateral amygdala damage, SM, using a short movie (Heider&Simmel, 1944) which shows 3 simple geometric shapes moving around a stationary rectangle.  Almost all normal observers of this movie describe it in anthropomorphic terms, with references to gender, agency, emotions, relationships, and personality traits (e.g., the big triangle is a mean, angry bully who tries to trap the defenseless circle).  In contrast, SM described the movie in strikingly inanimate terms, lacking not just emotion words (a deficit consistent with her well-characterized emotion recognition deficits) but also words like gendered pronouns, verbs indicating interactions, and nouns related to social roles.  However, she answered some, but not all, specific questions regarding the movie’s content fairly normally (e.g., she called the big triangle a bully when asked what it was like).  We interpreted this to mean that (1) SM fails to recognize the emotional salience of certain events in the movie, and thus (2) doesn’t construct a rich, social narrative of the movie, despite picking up on certain specific aspects of social meaning.

This interpretation is consistent with another result using the same stimulus and task:

Heberlein AS, Pennebaker JW, Tranel D, Adolphs R (2003) Effects of damage to right-hemisphere brain structures on spontaneous emotional and social judgments.  Political Psychology 24(4): 705-726.
A group of subjects with damage to right somatosensory cortex, a region of the brain which not only represents body sensations but has also been established as an important region for recognizing others’ emotion from faces, vocal prosody, and body movements (see below), described the Heider&Simmel movie using significantly fewer emotion words than did matched controls.  This is notable because the kinds of tasks normally used to test emotion recognition involve asking subjects to look at (or listen to) clear photographs, movies, or audio clips of emotional expressions and to label these (usually choosing from a list of emotion words) or rate them on scales.  Real-life emotion recognition doesn’t work anything like this; there are all kinds of ambiguous stimuli which are nonetheless interpreted emotionally.  So it’s interesting that the same neural regions (right somatosensory cortex and amygdala) are critical for emotion recognition both in the very focused contexts of rating/labeling and in the spontaneous, free-form context of describing the interactions of animated shapes.

Heberlein AS, Adolphs R, Tranel D, Damasio H. (2004) Cortical regions for judgments of emotions and personality traits from pointlight walkers. Journal of Cognitive Neuroscience. 16: 1143-1158.
We examined differences in regions of brain damage associated with emotion vs. personality trait judgments from the same stimuli.  We used a dynamic stimulus called point-light walkers, created by attaching small lights to a person’s major joints and head, and filming this person moving in the dark.  Examples can be viewed here, here, and here (please note that all movies are copyrighted but can be used, with appropriate citation, for research or educational purposes).  Overlapping sets of point-light walkers were used in an emotion labeling task and in a trait-rating task, and the brain lesions of all impaired (i.e., abnormal relative to a control group) subjects were mapped onto a common reference brain.  This lesion overlap technique, dubbed “dysfunctional imaging” by Lesley Fellows, allows a visualization of neural regions or structures which, damage to which is associated with impaired performance on some task.  Consistent with prior studies by Adolphs and colleagues, emotion recognition from point-light walkers depended on right somatosensory cortices (in the postcentral and supramarginal gyri).  In contrast, attribution of traits such as extraversion depended on regions in left inferior prefrontal cortex.

This double dissociation was upheld in a functional imaging study, performed in collaboration with Rebecca Saxe:

Heberlein AS, Saxe RR. (2005) Dissociation between emotion and personality judgments: Convergent evidence from functional neuroimaging.  NeuroImage.  28:770-777.
Neurologically healthy participants rated how well emotion terms or personality trait terms described each of the same set of point-light walkers.  We drew regions of interest on slices of each fMRI participant’s brain based on the lesion overlaps found in the previous study. There was greater activity in the right somatosensory region for emotion vs. personality trait judgments, with the opposite pattern in the left prefrontal region.

Atkinson,A., Heberlein AS, Adolphs R. (in press).  Spared ability to recognise fear from static and moving whole-body cues following bilateral amygdala damage.  Neuropsychologia.
(with Tony Atkinson and Ralph Adolphs)
We asked whether the same neural substrates, specifically the amygdala, play critical roles in emotion recognition from facial expressions vs. from other nonverbal cues.  To that end, we tested 2 patients with bilateral damage to the amygdala, a structure known to be important for recognition of emotions, especially fear, from facial expressions.  They were asked to rate or label photographs of emotional body postures with the faces blurred, as well as short videos depicting point-light and full-light body movement.   Both subjects recognized whole-body cues for all basic emotions tested, including fear, at normal levels, indicating that the amygdala is not critical for whole-body fear recognition.

 Heberlein AS,  Padon AA, Gillihan SJ, Farah MJ, Fellows LK (in press).  Facial emotion recognition impairments consequent to ventromedial frontal lobe damage.  Journal of Cognitive Neuroscience.

Farah MJ, Heberlein AS. (2007). Personhood and neuroscience:  Naturalizing or nihilating?  American Journal of Bioethics. 7(1): 37-48.
Farah MJ, Heberlein AS. (2007).  Response to Open Peer Commentaries on “Personhood and neuroscience: Naturalizing or nihilating?”: Getting Personal. American Journal of Bioethics. 7(1): W1-W4.
In a target article followed by multiple commentaries and our response, Martha Farah and I explored bioethical implications of scientific findings suggesting that we use heuristics to make personhood judgments (and thus that our intuitions regarding who has a mind, or other personhood-relevant features, may not be trustworthy.  One compelling demonstration of this is seen in the famous 2-minute movie constructed by Heider&Simmel (1944), depicting the movements of 3 simple geometric objects.  Normal subjects cannot seem to stop themselves from seeing and describing this movie as the intentional actions of feeling, thinking individuals.  We discuss various definitional criteria for personhood in light of such data.

Fellows LK, Heberlein AS, Morales DA, Shivde G, Waller S, Wu D.  (2005)  Method matters:  An empirical study of impact in cognitive neuroscience.  Journal of Cognitive Neuroscience.  17: 850-858.
All of the post-docs at the Center for Cognitive Neuroscience at the University of Pennsylvania in 2003-2004 worked on this project, a collaborative outgrowth of our biweekly lunch series.  We were interested in the ways cognitive neuroscience uses-- or fails to use-- evidence from convergent methods such as lesion/neuropsychology vs. functional imaging studies, given the inferential strengths and weaknesses of each method.  This led us to examine citation rates and patterns over a decade of cognitive neuroscience studies, which showed that (1) functional neuroimaging papers were cited much more often than lesion papers over this time period; (2) largely due to disproportionate appearances in high-impact journals; and (3) functional imaging papers tending to cite other functionoal imaging papers, but not lesion papers (while the reverse was much less true).  We concluded with suggestions for fostering better use of convergent data.

home