Schilbach et al. argue that economic scenarios may not recreate the interaction dynamics of everyday-life social encounters, on account of their implicit view of the subject as a simple recipient of information. Indeed, economic scenarios are poor stand-ins for natural interactions. A mature social neuroscience inevitably requires dynamic, “second-person” paradigms. However, some aspects of social cognition are best studied in controlled, restricted scenarios, where the subject is anonymous and unobserved, as the presence of another human presents as many interpretational problems as the absence of one. We propose that in some cases, controlled, first-person economic scenarios are still useful for a mature study of social interaction. These scenarios enable measures of individual differences and provide a window into basic brain-behavior relationships which can serve as a rudimentary “glossary” for decoding the complex datasets posed by second-person paradigms.
Experimental paradigms in social studies exist somewhere between two extremes: On one, we have controlled, observer-based protocols to study phenomena like empathy for pain or emotional reactivity. Individual subjects are exposed to stimuli which are ordered and controlled in order to group features of interest. These are then assigned to contrasts in order to isolate correlates of those features within our dataset of dependent variables. This allows us to abstract across conditions, making it possible to extrapolate and generalize. On the other extreme, we have observational studies of social phenomena in vivo, which assume that we cannot examine components of social scenarios in isolation without altering them beyond recognition. The first extreme is unrealistic, by assuming that there is such a thing as context-independent behavior. The second lacks the ability to say anything about anything other than its subject, if we cannot abstract and extrapolate beyond its context. In short, the first lacks validity, while the second lacks generalizability.
A second-person neuroscience may, in avoiding the problems inherent to the first extreme, stray too close to the faults of the latter: Namely, if we are to study interactions, how do we categorize them without somehow restricting them, and, most importantly here, how do we interpret data derived from interactions? Beyond inter-subject correlations such as those observed in hyper-scanning, what can we say about the neural mechanisms of interaction? To give an example by analogy, if we are studying a pair of linked oscillators, we are aware that their behavior in isolation may say little about their behavior in interaction. However, when in interaction, we can say nothing more than “they are in sync; they are decoupled, etc.” We can still say very little about how their internal architecture produces the observed behavior. But what if we could group varieties of oscillators according to their internal properties? Then, we might observe that, for example, AB, BB, and AA dyads tend to coalesce into significantly different stable states. Bringing the analogy back to social neuroscience, we propose that a study of individual differences rooted in internal structure and function can best link interaction-level data with the internal properties of the interactive agents.
The study of individual differences in empathy and prosocial behavior generally falls into two camps: Neurobiological studies have proposed “low-level” neural mechanisms for empathy based on mirroring and simulation (Carr et al. Reference Carr, Iacoboni, Dubeau, Mazziotta and Lenzi2003; Singer et al. Reference Singer, Seymour, O'Doherty, Kaube, Dolan and Frith2004). On the other hand, studies of prosocial behavior have focused on the role of empathy in our decision-making, primarily through observational studies of prosocial behavior, and economic studies of sharing and fairness using games (Fehr & Camerer Reference Fehr and Camerer2007). Each of these fields has historically measured individual differences using questionnaires. However, a handful of studies in the last few years have at last attempted to bridge the gap, foregoing questionnaires to directly relate differences in brain activity to differences in behavior. This sort of study, in our view, best exemplifies the utility of controlled economic scenarios. Our research, for example, attempts to correlate individual differences in neural correlates of mirroring with behavior in the Dictator game. Rather than attempting to mimic natural social interactions, we attempt to isolate neural biomarkers of noncompliant prosocial behavior. Subjects are asked to share a portion of a sum of money with virtual profiles which represent real people in the community who will actually receive the money. By assuring subjects that they are anonymous and unobserved, we can isolate neural correlates of prosocial inclination or empathic concern, as we are not confounding their behavior or cognition with the presence of another in interaction. Because, in point of fact, it is in those very scenarios where we are unobserved, and anonymous, yet still aid others, that we can be said to act without concerns for our reputation or due to social pressure. Thus, to best measure differences in brain activity, connectivity, and structure which map onto variability in controlled, noncompliant altruistic behavior, subjects should not interact. We believe that the results of this sort of study can allow for neural biomarkers of prosocial behavior which can aid in the interpretation of behavior patterns within interaction.
We agree that interactions between multiple subjects may not be reducible to the characteristics of their participants. Dyads, triads, and other group configurations may one day be the basic unit of social neuroscience. We may one day refer to “group states” in social neuroscience, as a unit of information separate from and incommensurable with the states of the individuals within them. This parallels the study of quantum mechanics and complexity theory, which require us to consider the behavior of systems holistically, defying the reductionist notion that group states are reducible to the sum of their parts. However, the volume of data that we can expect from studies of social interaction, and the intricate complexity of relating this data back to mechanistic models of brain function, suggests that it may be useful to parameterize the interactive agents. This can be done by measuring individual differences based on internal functions, even if the paradigms we use to do this sometimes bear little resemblance to natural scenarios.
You have
Access
Schilbach et al. argue that economic scenarios may not recreate the interaction dynamics of everyday-life social encounters, on account of their implicit view of the subject as a simple recipient of information. Indeed, economic scenarios are poor stand-ins for natural interactions. A mature social neuroscience inevitably requires dynamic, “second-person” paradigms. However, some aspects of social cognition are best studied in controlled, restricted scenarios, where the subject is anonymous and unobserved, as the presence of another human presents as many interpretational problems as the absence of one. We propose that in some cases, controlled, first-person economic scenarios are still useful for a mature study of social interaction. These scenarios enable measures of individual differences and provide a window into basic brain-behavior relationships which can serve as a rudimentary “glossary” for decoding the complex datasets posed by second-person paradigms.
Experimental paradigms in social studies exist somewhere between two extremes: On one, we have controlled, observer-based protocols to study phenomena like empathy for pain or emotional reactivity. Individual subjects are exposed to stimuli which are ordered and controlled in order to group features of interest. These are then assigned to contrasts in order to isolate correlates of those features within our dataset of dependent variables. This allows us to abstract across conditions, making it possible to extrapolate and generalize. On the other extreme, we have observational studies of social phenomena in vivo, which assume that we cannot examine components of social scenarios in isolation without altering them beyond recognition. The first extreme is unrealistic, by assuming that there is such a thing as context-independent behavior. The second lacks the ability to say anything about anything other than its subject, if we cannot abstract and extrapolate beyond its context. In short, the first lacks validity, while the second lacks generalizability.
A second-person neuroscience may, in avoiding the problems inherent to the first extreme, stray too close to the faults of the latter: Namely, if we are to study interactions, how do we categorize them without somehow restricting them, and, most importantly here, how do we interpret data derived from interactions? Beyond inter-subject correlations such as those observed in hyper-scanning, what can we say about the neural mechanisms of interaction? To give an example by analogy, if we are studying a pair of linked oscillators, we are aware that their behavior in isolation may say little about their behavior in interaction. However, when in interaction, we can say nothing more than “they are in sync; they are decoupled, etc.” We can still say very little about how their internal architecture produces the observed behavior. But what if we could group varieties of oscillators according to their internal properties? Then, we might observe that, for example, AB, BB, and AA dyads tend to coalesce into significantly different stable states. Bringing the analogy back to social neuroscience, we propose that a study of individual differences rooted in internal structure and function can best link interaction-level data with the internal properties of the interactive agents.
The study of individual differences in empathy and prosocial behavior generally falls into two camps: Neurobiological studies have proposed “low-level” neural mechanisms for empathy based on mirroring and simulation (Carr et al. Reference Carr, Iacoboni, Dubeau, Mazziotta and Lenzi2003; Singer et al. Reference Singer, Seymour, O'Doherty, Kaube, Dolan and Frith2004). On the other hand, studies of prosocial behavior have focused on the role of empathy in our decision-making, primarily through observational studies of prosocial behavior, and economic studies of sharing and fairness using games (Fehr & Camerer Reference Fehr and Camerer2007). Each of these fields has historically measured individual differences using questionnaires. However, a handful of studies in the last few years have at last attempted to bridge the gap, foregoing questionnaires to directly relate differences in brain activity to differences in behavior. This sort of study, in our view, best exemplifies the utility of controlled economic scenarios. Our research, for example, attempts to correlate individual differences in neural correlates of mirroring with behavior in the Dictator game. Rather than attempting to mimic natural social interactions, we attempt to isolate neural biomarkers of noncompliant prosocial behavior. Subjects are asked to share a portion of a sum of money with virtual profiles which represent real people in the community who will actually receive the money. By assuring subjects that they are anonymous and unobserved, we can isolate neural correlates of prosocial inclination or empathic concern, as we are not confounding their behavior or cognition with the presence of another in interaction. Because, in point of fact, it is in those very scenarios where we are unobserved, and anonymous, yet still aid others, that we can be said to act without concerns for our reputation or due to social pressure. Thus, to best measure differences in brain activity, connectivity, and structure which map onto variability in controlled, noncompliant altruistic behavior, subjects should not interact. We believe that the results of this sort of study can allow for neural biomarkers of prosocial behavior which can aid in the interpretation of behavior patterns within interaction.
We agree that interactions between multiple subjects may not be reducible to the characteristics of their participants. Dyads, triads, and other group configurations may one day be the basic unit of social neuroscience. We may one day refer to “group states” in social neuroscience, as a unit of information separate from and incommensurable with the states of the individuals within them. This parallels the study of quantum mechanics and complexity theory, which require us to consider the behavior of systems holistically, defying the reductionist notion that group states are reducible to the sum of their parts. However, the volume of data that we can expect from studies of social interaction, and the intricate complexity of relating this data back to mechanistic models of brain function, suggests that it may be useful to parameterize the interactive agents. This can be done by measuring individual differences based on internal functions, even if the paradigms we use to do this sometimes bear little resemblance to natural scenarios.