Rethinking Psychology by Applying Engineering to the Study of Reward/Aversion
A tremendous amount of progress has occurred in our understanding of motivated behavior since Acquinas first speculated about intention and Spinoza hypothesized a continuum between pain and pleasure as the stimulus behind intention. Despite this progress, and some models of reward/aversion processing leading to Nobel Prize recognition, we still lack an overarching, quantitative platform for describing approach and avoidance decisions. This problem plays into a larger issue with our descriptions of behavior and the fact there still is no chart of psychological functions with the same quantitative impact as the chart of biochemical pathways (e.g., a “Krebs cycle” linking signal detection theory, Ebbinghaus memory functions and reward/aversion processing). It could be argued that until such a “hard” perspective toward behavior develops, most behavioral descriptions are “fuzzy”, “soft”, and difficult to test, leading to a significant problem with design efforts that seek to incorporate quantitative patterns of behavior into the engineering of consumer products and solutions to significant world problems with mental illness.
To address this challenge, we considered the dictates of Feynman regarding lawfulness, and applied an iterative modeling approach to approach/avoidance behavior to see if we could observe patterns that were consistent with these dictates, and encoded critical features of older models of reward/aversion processing. This process led to the discovery of patterns in approach/avoidance that were algorithmic/mathematical, recurrent across experiments, nontrivial and robust to the addition of noise, and scalable. These patterns in approach/avoidance [referred to as relative preference (RP)] represented a merger of critical features of other reward/aversion theories including matching (reinforcement learning), alliesthesia (hedonic deficit theory), mean-variance theory, and prospect theory. Intriguingly, RPT could be used to derive prospect theory and matching. RPT also was the only construct for reward/aversion based on variables within the individual, and thus consistent with agency as opposed to being a system calibrating internal experience to external measures. It has further proven to be a construct that could be integrated with other quantitative constructs in psychology, such as signal detection parameters around a vigilance paradigm. Other work has shown RPT scales from behavior to fMRI signal, and can model important aspects of psychopathology that were associated with alterations in brain structure.
Going forward, a number of issues still remain regarding RPT and its capacity to serve as an organizing theory for approach/avoidance. First we need to unpack the potentially opponent relationship between approach and avoidance, to better understand its effect on the trade-off between them. Second, we need to assess how power law patterns in fMRI BOLD signal might relate to MRI-based measures of brain structure such as gray matter density and behavioral measures encoded by RPT. Lastly, we need to determine how RPT can be integrated with signal detection measures of other types of attention, and quantitative metrics of memory to produce a more integrated foundation for psychology. Such integration might also facilitate consideration of how patterns in preference-based decision-making might be combined with free energy models of behavior studied by other Segal 2.0 faculty.
Hans C. Breiter, M.D., Professor of Psychiatry and Behavioral Science, is a psychiatrist and neuroscientist at Northwestern University Feinberg School of Medicine (NUFSM) and Massachusetts General Hospital (MGH). Trained in mathematical logic before medical training, he completed a psychiatry residency and five post-doctoral fellowships in neuroanatomy, functional magnetic resonance imaging (fMRI), experimental psychology, pharmacokinetics/dynamics, and addiction neuroscience. He has been PI of the MGH Phenotype Genotype Project on Addiction and Mood Disorder since 2003, an interdisciplinary multi-center project involving more than 80 investigators integrating experimental psychology, multi-modal neuroimaging, and genetics. He has directed the Motivation and Emotion Neuroscience Collaboration (MENC) since 1999, and in 2008, became Co-Director of the MGH Translational Center for Prescription Drug Abuse. He is a founding investigator in the Brain Architecture Project centered at Cold Spring Harbor Laboratories. In 2011, he became Scientific Director of the Warren Wright Adolescent Center at NUFSM, and in 2012, he directed the development of the Collaborative Neuromarketing Group at Northwestern University. His publications have generated 10,000 citations.
His research career started with guiding the team that first built an fMRI analysis pipeline between 1992-94 (see David et al., 1994), allowing psychiatric applications of fMRI (Breiter et al., 1993; 1995; 1996), for which he was awarded the Klerman Award in 1996 by NARSAD. As part of this development work, he was a primary investigator with the team that produced the first fMRI and EEG integration (Huang-Hellinger et al., 1996). He also was part of the first teams to apply these integrated methods for fMRI analysis to quantitative cognitive science research (Cohen et al., 1996; Seidman et al., 1998). Work published in Neuron in 1996 and 1997 localized human reward circuitry, and stimulated development of the field of reward/aversion neuroimaging, which is now a salient focus of the Organization for Human Brain Mapping, and the Society for Neuroeconomics (Breiter et al., 1996, 1997; Breiter & Rosen, 1999). Work performed with Danny Kahneman and Peter Shizgal between 1997-2001 demonstrated that aspects of Kahneman’s prospect theory accurately modeled human reward/aversion processing, and was one of the first studies in neuroeconomics using neuroimaging (Breiter et al., 2001). Kahneman won the Nobel Prize in Economics in 2002 for developing prospect theory.
Concurrent work at that time indicated that a common circuitry processed reward and aversion stimuli, consistent with Spinoza’s thesis of a continuum between pain and pleasure (Becerra et al., 2001); other work tied aesthetic processes to reward/aversion systems (Aharon et al., 2001), and showed that a common set of brain regions processed drugs, money, and social stimuli. These studies with pain and aesthetic stimuli were among the first publications using neuroimaging to address philosophical questions. The consolidated implications of these findings were woven into a theoretical background for judgment and decision-making in health and psychiatric illness between 1998 and 2003 (Breiter & Rosen, 1999; Breiter and Gasic, 2004). Through the Phenotype Genotype Project (2003-2009; see Breiter et al., 2006), his team produced some of the first associations between microeconomic measures and genes implicated in depression and addiction in 2008 (Perlis et al., 2008; Gasic et al., 2009), and developed a model of reward/aversion function that synthesizes Kahneman’s prospect theory with two other reward/aversion theories (Kim et al., 2010). This work, called Relative Preference Theory (RPT), identified recurrent, robust, and scalable patterns in human preference-based decision-making. RPT appears to be lawful, and has been connected with brain reward circuitry (Makris et al., 2008) plus genes modulating these circuits. RPT has also demonstrated quantitative phenotypes for cocaine dependence and alcohol dependence, which can be integrated with multimodal imaging measures to raise mechanistic hypotheses regarding these disorders. Ongoing work at Northwestern is evaluating how RPT may be connected to signal detection theory and Ebbinghaus memory functions.