Neuroeconomics

Genetic Expression may Determine Handedness

There is a nice review in Nature Reviews Neuroscience on the mechanism that causes asymmetry in the brain. One of the theories discussed is that molecules produced in the embryonic brain that induce cell specialization are distributed asymmetrically in the hemispheres, which starts a chain reaction leading to hemispheric specialization. Competing theories have suggested fetal position, or development that occurs postpartum.

Handedness is interesting to neuroeconomists and others because imaging studies frequently exclude left-handed individuals because activation patterns in left-handed individuals are inconsistent even among lefties, and don't correlate well with data obtained from right-handed individuals.

Almost of equal interest in this review is the fact that hemispheric specialization may be influenced by the "Sonic Hedgehog" gene. I'm glad that (a) I have a Sonic Hedgehog gene and (b) such a gene exists in the first place.

Multidimensional decoding of mental states

    Conventional analysis of neuroimaging seeks to determined how a particular perceptual or cognitive state is encoded in the brain by determining what regions of the brain are involved in a task. The current methodological approach is the use of location-based analyses of brain activity. These methods consist of multiple repetitions of measurement of brain activity from thousands of locations in the brain, whereby each location is analyzed separately.  In a recent study in Nature Neuroscience, spatially distributed pattern recognition has been offered as an alternative to conventional forms of analysis. 

    The core distinction between the two methodological types is the use of univariate versus multivariate analysis of the imaging data. The univariate analysis is used in conventional neuroimaging studies focusing on comparing multiple mental states at the same location, which looks for  significant differences of averages between activity across all task in one condition and activity across all time points in a second condition. Multivariate analysis focuses on accumulation and aggregation of fine-grained, spatially distributed information, which use decoding-based approaches that could allow for quasi-online estimates of a person's perceptual and cognitive state.

    The decoded-based approaches presented by Hayes and Rees apply pattern recognition techniques to neuroimaging data, which intend to decode a person's mental state through learning to recognize spatial patterns of brain activity associated with mental states. Given that this approach takes account of full spatial pattern of activity, it shows that significantly more information is encoded in fMRI signals that was previously considered.

    According to the authors, these approaches can be use to address the question of how perceptual and cognitive state information is encoded in the brain, which has been partially achieved by the use of pattern-based decoding in revealing the principles of underlying representation of ventral visual pathways. It can also reveal the conscious and unconscious representations of individual features. Although they seem to provide a very acute way of determining what types of information are represented in a spatially distributed pattern of brain responses given the existing technology, they seem to be limited and narrow in their use for general applications including the ability to generalize across individuals and different cognitive and perceptual states.

   Furthermore, these techniques put forth the ability of revealing personal information at either a conscious or unconscious level without the necessity of knowledge or consent, thus raising a very important ethical concern in the form of mental privacy.

   

Hyperscanning

I just got back from Baylor College of Medicine where Read Montague and his group at the Human Neuroimaging Laboratory. One of their big projects is Hyperscanning, i.e., the ability to image two or more brains as they interact concurrently. What exactly does this mean for neuroeconomics? In economics, we think of multiple minds engaged in distributed computations in order to solve either production or exchange decision problems. But do we need more than one scanner? It seems that when two or more minds interact they do so through their sensory systems. So in principle we can look at just one brain as it receives sensory information that reflects the results of another brains’ computation. Is this enough? Since brains are wired for social interchange it may not be enough. Take for example theory-of-mind, i.e., the ability to infer intentions about another person. Can we infer all intentions (this doesn’t seem to be the case), or just some intentions that occur when two brains are engaged in particular computations? If this is the case, then without hyperscanning it would be very difficult to test inferences about what these joint computations may be.