Click here for a list of lab publications organized by research topic

Background. Most of the research in my lab involves a social cognitive neuroscience (SCN) approach that I developed with Kevin Ochsner when we were graduate students at Harvard University. Social cognitive neuroscience focuses on how the human brain carries out social information processing. Practically speaking, this means that we use functional neuroimaging (fMRI) and neuropsychology to test new hypotheses regarding social cognition or old questions whose answers continue to elude us.

It is usually taken as given that automatic and spontaneous affective processes are more resistant to disruption than thought. In other words, it is easier to disrupt our thoughts ("now I'm going to think about fire hydrants, instead of bananas") than our feelings ("now I'm going to be happy instead of sad"). I have theorized that sometimes thoughts do disrupt or turn down the intensity of our automatic affective reactions because of the evolutionary functions of (and relationship between) the systems in the brain that produce automatic affect and thought.

One system in the brain (the 'X-system' for the 'x' in reflexive) responds automatically to well-learned affective contingencies and spontaneously sets appropriate behavior and cognitive responses in motion. The X-system consists of the amygdala and basal ganglia which are optimally tuned to negative and positive cues, respectively (the X-system also includes lateral temporal cortex which stores social and non-social semantic associations which form much of our background knowledge about the world). As long as the X-system can match external cues to its existing repetoire of affective representations, the individual can move through the world seamlessly without exerting much mental effort. However, when the situation is novel or when existingresponses are contextually inappropriate, the X-system cannot function adaptively. The C-system (for the 'c' in reflective) can exert control in these instances and exhibit contextually-sensitive flexibility that is absent from X-system functioning. The C-system consists of the anterior cingulate cortex, prefrontal cortex, and the medial temporal lobe. The C-system, while flexible, is quite fragile. The C-system is largely responsible for conscious thought and as we all know, it is virtually impossible to have two thoughts at the same time. Trying to remember a new phone number and a new web address at the same time will likely lead to neither being successfully remembered.

If the C-system evolved to come online and save the day at precisely those moments that the X-system is getting us into trouble, it would be adaptive for the C-system to disrupt the functioning of the X-system whenever it is called upon. In this way, the brain can avoid having two competing responses fighting it out. The last thing the fragile C-system needs is to be fighting directly against the continued misguided output of the X-system. By way of neural connections from dorsolateral prefrontal cortex and the anterior cingulate cortex to the basolateral nucleus of the amygdala, the C-system may disrupt the X-system whenever it is called upon. One interesting implication of this hypothesis is that in the modern world, we may end up turning off our automatic affective sensitivities even when these X-system processes are not getting us into trouble. Today, we use the C-system extensively even when the X-system is on track because the use of language engages the C-system. So it may be the case that whenever we think and use language, we are less sensitive to affective cues in the environment.

We have now shown in both neuroimaging research of automatic affective responses to racial outgroups and behavioral studies with a subliminal version of the mere exposure paradigm, that minimal engagement of the C-system is sufficient to disrupt genuinely automatic processes in the X-system (Lieberman, 2002; Lieberman, & Jarcho, 2002; Lieberman, Hariri, Jarcho, Eisenberger, & Bookheimer, 2005).

back to the top

The Neural Bases of Automaticity and Control in Social Inference Processes

Background. Much of my current work focuses on how automatic and controlled processes interrelate in the human brain to produce social inference (Lieberman, Gaunt, Gilbert, & Trope, Advances v. 34). Automatic processes tend to operate without awareness, effort, intention, volition, or interference with other ongoing processes. Controlled processes tend to have more of these features (they require effort, etc). Simple over-rehearsed behaviors are very automatic, like stepping on the break when you see a stop sign. Rehearsing a phone number as you go from the phone book to the phone is a prototypical controlled process. But what about something like typing? You consciously control what words to type and yet at the same time your fingers seem to be doing their own thing automatically. Here is where things get interesting, when there are both automatic and controlled processes occurring simultaneously. How do these processes interface and what happens when you try to exert more or less control?

Implicit Learning & Nonverbal Communication. My dissertation (Lieberman & Rosenthal, 2001) examined the automatic and controlled components of nonverbal decoding in introverts and extraverts. Nonverbal decoding involves making sense of the meaning of nonlinguistic cues given off by interaction partners. Given that nonverbal cues carry so much of the affective and social evaluative meaning that contextual what we say to each other, accurate nonverbal decoding is critical to social inference processes. So how is it that we learn how to interpret these series of subtle nonverbal cues given that the rules are never explicitly taught? Implicit learning processes, studied by cognitive psychologists (Reber, 1967; Knowlton & Squire, 1996), may produce exactly the kind of learning and functional dynamics associated with nonverbal communication (Lieberman, 2000). Barbara Knowlton and I have recently finished an fMRI study of implicit learning (Lieberman, Chang, Chiao, Bookheimer, & Knowlton, 2002) in which we found that the basal ganglia was unique involved in discriminating stimuli that followed very subtle sequential rules from stimuli that did not. In future studies, we hope to show that the basal ganglia is also involved in nonverbal decoding of moving faces, as opposed to still images.

Relationship Attributions. Alan Fiske has demonstrated that across cultures when an individual sees other people interacting, they automatically encode those relationships into discrete categories (Communal Sharing, Authority Ranking, Equality Matching, & Market Pricing). We have recently finished an fMRI study in which we examined the neural correlates of these relationship attributions when participants were viewing communal, authority, or nonsocial videoclips (Iacoboni, Lieberman, Knowlton, & Fiske, 2002).

Causal Inference Processes. Since the seminal work of Hal Kelley and Ned Jones on attribution, the use of causal inference strategies has been a central part of social inference research. Keith Holyoak and I have recently looked at the neural correlates of causal inference as compared with closely matched associative inference tasks (Lieberman, Sellner, Satpute, Waldman, & Holyoak, in prep). We have found that even when the tasks are matched for effort, there is an increase in activity in dorsolateral prefrontal cortex associated with the causal, but not with the associative, task suggesting that causal inference involves qualitatively different computations than associative inference as suggested in my model of X- and C-system processes (Lieberman, Gaunt, Gilbert, & Trope, Advances v.34). In other work, David Sherman and I have conducted a series of studies looking at the extent to which causal inferences are made automatically from correlational data.

Self-Schemas. The fact that people come to represent information about their self-attributes very efficiently in self-schemas has been a mainstay of social cognition since its inception. Cognitive neuroscience however has little to say about how self-schematic information is represented in the brain. In other words, what kinds of different neurocognitive processes are involved when making self-attributions in domains that we have thought about repeatedly and are important to our identity, as compared to making self-attributions in other domains. Following Markus (1977), we have recently finished an fMRI study in which college soccer players and improvisational actors made self-attributions in both domains (soccer and improv acting). Thus far we have found changes in the temporal pole, superior temporal sulcus, hippocampus, basal ganglia and frontal poles associated with domain expertise (Lieberman, Jarcho, & Satpute, in progress).

Expression vs. Identity Recognition of Other Races. Jennifer Pfeifer, Russ Poldrack and I have begun investigating the social cognitive and neurocognitive components of the same -race advantage that people have in recognizing the expression and identity of members of their own race compared with members of other races. In particular, we are testing the hypothesis that perceptual expertise moderates the identity bias while implicit attitudes moderates the expression bias.

Attribution & Culture. Cross cultural research generally assumes that individuals from interdependent cultures show a diminished correspondence bias in their attributions. There are two possible reasons for this difference from our culture. Individuals from interdependent cultures might be automatically predisposed to take situational information into account. Alternatively, they might make a greater conscious effort to correct their initial attributions to take the situation into account. We have used the cognitive busyness paradigm with subjects from Japan and America and have found no cross cultural differences in the automatic component of the attribution process suggesting that an attenuated correspondence bias is the result of effortful process (Lieberman, Jarcho, & Gilbert, in prep)
 

back to the top

Personality theories have long tried to use self-report scales, designed to index hypothesized individual differences in neural reactivity, in order to predict cognition, behavior, and affect. With the advent of functional magnetic resonance imaging (fMRI) it is now possible to quantify these differences much more directly than we have been previously able with personality scales.  We provide evidence from psychiatric disorders, neurophysiology, and computational neuroscience that together suggest the anterior cingulate cortex (ACC) may play an important role in neuroticism and the social cognitive outcomes conceptually associated with neuroticism.  Using a social cognitive neuroscience approach (Lieberman, 2000; Ochsner & Lieberman, 2001), we propose to look at the ways in which individual differences in ACC reactivity shape social cognition and how social factors, in turn, can shape ACC reactivity.  In particular, Naomi Eisenberger and I have found that participants with more sensitive rostral ACC's tend to be more accurate in their self-appraisals of physiological arousal across several time points following vigorous exercise.
 

Predicting, Experiencing, and Constructing Positive Experiences

This research involves a number of loosely related areas of study, some of which involve cognitive neuroscience methodologies and others which do not. Along with Dan Gilbert and Tim Wilson, we are examining how and why individuals mispredict how resilient they will be when they are the victim of an interpersonal transgression (Gilbert, Lieberman, & Wilson, 2002). In general, people expect to stay mad longer than they do because they have a psychological immune system. We have also collected data with retrograde amnesic patients in a cognitive dissonance paradigm that suggests that this psychological immune system can operate unconsciously (Lieberman, Ochsner, Gilbert, & Schacter, 2001). We have also begun work on an fMRI study to look at the neural bases of placebo effects (Jarcho & Lieberman, in progress).

back to the top