Share this article:
The New Psychology: Simulations
This is the third in my four-part blog about The New Psychology. The natural science explanations provided by New Psychology require new tools. The ability to simulate neural network events is a critical new tool. I begin this blog with a description and justification of simulation as an important New Psychology research tool. Then I illustrate some of the contributions that have been made using simulations. I end with conclusions.
I understand that most psychologists, especially clinical psychologists, are not interested in acquiring and will not acquire the skills necessary to conduct neural network simulations any more than they will acquire the skills necessary to operate a brain scanner and conduct all post processing data analyses. Instead, they will collaborate with colleagues who have these skills. Psychology departments that offer doctoral training in psychometrics/quantitative psychology may begin to train the mathematical and computer skills necessary to do these simulations. Graduates of these programs will then be able to consult with other psychologists in the same way that they would consult with other psychologists who wish to conduct complex statistical analyses. These psychologists will become part of the interdisciplinary team and a co-author on any resulting publications.
But, future psychologists will be increasingly expected to be conversant with the basic features of simulations just as they will be increasingly expected to be conversant with brain imaging. Future psychologists will need to understand why simulations are important and what they can do.
In my book (Tryon, 2014), Cognitive Neuroscience and Psychotherapy: Network Principles for a Unified Theory and previous blogs, I have consistently criticized what I now refer to as Old Psychology for failing to provide causal mechanism information. The required causal mechanism information consists of many steps that are too numerous and complex to be adequately described verbally. New methods are required to track and study these complex interactions. Computer simulation is a good way to do this.
Parallel distributed processing (PDP) connectionist neural network (CNN) models are complex systems. Their sensitivity to initial conditions and nonlinear dynamics requires simulations to track how these systems develop their “adult” capabilities.
A corollary point is that natural science requires that all hypotheses be falsifiable. This means that they must be stated in a way that can be contradicted by observable facts. The problem here is that verbal descriptions of PDP-CNN models and how they operate are not falsifiable because they do not result in specific outcomes that can be compared with, and possibly contradicted by, data from real people. Simulations are required to translate neural network theories into specific outcomes that can be supported or contradicted by real data.
Experiments enable one to draw cause and effect conclusions. Experiments require clear definition of terms, control over how variables interact, and the ability to hold constant variables that might otherwise confound conclusions. Simulations satisfy all of these requirements. Hence, simulations enable rigorous and therefore persuasive experiments to be conducted.
Most psychologists do not conduct brain experiments but their explanations increasingly make reference to brain structures and functions. PDP-CNN models and their simulations provide a way for psychologists to experiment with brain-like models. Serious experiments are being conducted using computer simulations. The Dawn Blue Gene/P supercomputer built by IBM can simulate 1.617 billion neurons and 8.87 trillion synapses. A major problem with such tools is that they consume a great deal of power and run more slowly than real brains. These problems are solved by hardware simulations using neuromorphic chips. This can be done because the physical forces that drive electrons through transistors are the same forces that drive ions across cell membranes. Kwabena Boahen described his work with neuromorphic chips at the following link: http://www.youtube.com/watch?v=mC7Q-ix_0Po. The following link is to Boahen’s lab: http://www.stanford.edu/group/brainsinsilicon/.
In sum, simulation is a major new tool for the natural science study of neural network models. Simulations enable carefully controlled experiments to be conducted on complex neural networks concerning topics of psychological interest. I refer to these simulations as computational neuropsychology. These studies advance our understanding of how mind emerges from brain. These simulations integrate psychology and neuroscience in a cooperative and constructive way. There is no competition here between psychology and neuroscience nor is there any threat of replacement. The only thing that is getting replaced is Old Psychology by New Psychology and psychological science will be better off as a result. Most psychologists will collaborate with colleagues who have the required technical skills to conduct simulations like they presently do when advanced statistical analyses and/or brain imaging is involved.
Critical Features of Simulations
Simulations simplify. All neural network simulations are simplifications for at least the following two reasons. First, it is not possible to create a simulation that is as complex as the human brain. While this remains a very long-term goal, it is presently impossible. Second, it would not be helpful to begin with a models that is as complex as the human brain because then the model would not be any easier to study than the brain is. Biologists have long benefited from studying simple life forms as a way to better understand more complex life forms. Simulations therefore begin with simple models that contain basic and essential features.
Simulations simulate rather than reproduce the physical mechanisms that are involved in real neural networks. The Old Psychology criticism that PDP-CNN models sometimes use biologically implausible methods to simulate known neuroscience phenomena misunderstands that simulations simulate; they do not aim to reproduce exact biological mechanisms. Other sciences use simulations without this criticism. For example, astronomical simulations of supernova are not composed of star stuff nor do they explode. Meteorological simulations do not contain wind and water nor do they rain. Software simulations use mathematical equations to represent basic and essential features. Hence, it is ok that PDP-CNN models use mathematical methods such as the Back Propagation algorithm to simulate experience-dependent plasticity that modifies synaptic properties. It is the functional feature of synaptic modification caused by interaction of brain-like models with a simulated environment that is critical here.
Simulations animate. They bring explanations to life by animating the causal processes involved. Simulations are dynamic presentations. Simulations are like movies in that they capture and reveal how dynamic processes unfold over time.
Neural network simulations enable psychologists to study how psychological properties emerge from brain-like models. Simulations enable psychologists to empirically study the process of emergence rather than just speculate about it philosophically. This is a first step to understanding how mind emerges from brain.
The field of computational neuroscience simulates basic neuroscience phenomena that most psychologists have little interest in. The field of computational neuropsychology focuses on emergent phenomena such as personality and attitude formation and change that psychologists are quite interested in. I now provide a few examples of interesting psychological simulations.
Examples of Psychological Simulations
Read and Miller (2002), and especially Read et al. (2010), developed neural network models of personality that have at least two remarkable features. First, they integrate psychology and neuroscience by showing that neural network models can effectively simulate human behavior. Second, they integrate two major and seemingly incompatible approaches to the study of personality. I refer to the ideographic and nomothetic approaches to studying personality. The ideographic approach to personality studies individuals intensively over time. Freud is well known for this type of personality study. The nomothetic approach to personality studies groups of individuals mostly at a single point in time using a standardized test sometimes called a personality inventory. They use statistical methods to extract underlying factors. The Five Factor Model of personality is a good example of this approach. The neural network models created by Read and his colleagues allow for the simulation of any number of people from 1 to many and does so in a way that is fully consistent with neuroscience. Their simulations agree remarkably well with actual human data.
How do people make decisions? What dynamic processes are involved when people make decisions? The general answer to these questions provided by Old Psychology is that people make decisions using their mind and free will. It follows that only the person in question can explain why they decided as they did. The assumptions of mind and free will preclude the development of natural science explanations based on deterministic laws.
Neuroscience provides a natural science explanation. Gazzaniga (2011), in his book entitled Who’s in Charge? Free Will and the Science of the Brain, reported that consciousness does not result from any particular part of the brain but emerges from each of the great many neurons that are clustered into the network of neural networks that constitutes a brain. These neural networks function as a democracy. Their constituent neurons “vote” for or against every thought, feeling, or action in direct proportion to their level of activation. The majority rules as it does in every democracy. The majority neuronal “vote” is the brain’s decision. Our sense that we made a conscious decision is generated by the language centers in the left brain from one third of a second to ten seconds after the “vote” has occurred. Our language center, called the interpreter by Gazzaniga (2011), readily confabulates, and thereby creates, a personal narrative that fills in gaps with plausible gist to make a good story that “explains” how and why we came to the brain’s decision. Confabulation always occurs because our language centers do not have full access to all of the “voting” neurons. We access this highly synthesized personal narrative when we ask people why they did something, though in a particular way, or had a specific feeling. It is for these reasons that Nisbett and Wilson (1977) was justified in concluding long ago that people tell more than they can possibly know. This is a good example of why psychology needs to be constrained by neuroscience.
My book, Cognitive Neuroscience and Psychotherapy: Network Principles for a Unified Theory, discusses decision making as a constraint satisfaction process that is based on the well replicated finding that people are motivated towards consistency and away from dissonance. In Chapter 6, I present the work of Monroe and Read (2008) who proposed a general connectionist model of attitude formation and change as follows:
Briefly, we take the position that attitudes are best represented as networks of associated cognitions, with both positive and negative links among them, and that processing proceeds by the parallel spread of activation along those links. Moreover, long-term attitude change is the result of changes in the magnitude and direction of links among those cognitions. This formulation, a localist connectionist model, allows us to computationally model attitudes and, in turn, potentially allows us to identify fundamental properties that follow from such a configuration. We believe the existing research and evidence point in the direction of this formulation.
Monroe and Read (2008) reported findings from seven successful simulations. Their first simulation replicated the finding that attitudes become more extreme when people think about them; when they are more activated. Their second simulation demonstrated that the greater the interconnectedness of attitudes, the more they resist change, because greater connectivity results in more constraints. Strongly held beliefs and goals augment these constraints.
Considerable research demonstrates that cognitive abilities decline from early adulthood on. Li, Lindenberger, and Sikström (2001), Li, Naveh-Benjamin, and Lindenberger (2005), and Li, and Sikström (2002) discovered through simulation that these complex changes could be simulated by altering a single parameter that governs how much input is needed to activate a simulated neuron. Aging seems to require more input and that takes longer to occur. This explains why older people are slower and do less well than younger people on a wide variety of cognitive tasks.
Schizophrenia is major mental illness that for many years was thought to have a psychological origin. Psychologists once blamed schizophrenogenic mothers for the onset and development of schizophrenia (Neill, 1990). Investigators tried unsuccessfully for years to establish a psychological cause by attempting to distinguish patients diagnosed with schizophrenia from those diagnosed with brain disorders. These studies failed because schizophrenia is a brain disorder.
Insel (2010) noted that the number of prefrontal excitatory synapses begin at around 50% at birth, increase to a maximum of 100% around age 5 and progressively decrease to about 40% by age 20 where upon they remain relatively constant from then on. This decrease is the result of synaptic pruning.
Nature solved the problem of how to get the right genes into the right cells by putting all genes into every cell and activating only the right genes. Likewise, nature solved the problem of how to form the right neural circuits by over connecting brain neurons at birth and letting neurons that fire together to wire together by strengthening the synapses that connect them and cannibalizing unnecessary synapses to save energy. Brain development is like sculpting. The artist begins with a block of stone and chips away the stone that is not part of the desired result. The end product consists of stone where it needs to be and not in other places. For example, if the sculpture is of a horse then stone that does not represent a horse must be removed. Brain development is like this. Synapses that are not part of circuits that enable thinking, etc. are removed.
Synaptic pruning normally stops around age 20. Schizophrenia develops when synaptic pruning continues significantly beyond this age which is why the onset of schizophrenia typically occurs during the early twenties. Dhindsa and Goldstein (2016) and Skar et al. (2016) recently revealed genetic mechanisms discussed by Insel (2010) and Meyer-Lindenberg (2010) that explain how and why over pruning occurs in some individuals but not others.
Old Psychology cannot do much with this information except to add it to their list of biological factors. Old Psychology has no formal way to include this information into its explanations because it has no solution to the mind-body (brain) problem. New Psychology has already incorporated this information into connectionist models that simulate the effects of excessive pruning. I refer to simulations by Hoffman and Dobscha (1989) and Hoffman and McGlashan (1993, 1997) where connectionist models developed “loose associations” after connections were removed to simulate over pruning.
Simulations are a critical new tool of New Psychology that enables theories formulated in words to be translated into connectionist models that are trained and develop to where they generate results that can be falsified by comparing them with data from real people. Old Psychology cannot conduct such simulations because it lacks the required mechanism information. The ability to simulate psychological effects using brain-like models is revolutionary and is sufficient reason to adopt and endorse the New Psychology today.
Warren’s book, Cognitive Neuroscience and Psychotherapy: Network Principles for a Unified Theory is available for purchase on the Elsevier Store.
Use discount code “STC215” at checkout and save up to 30% on your very own copy.
About the Author
Warren W. Tryon received his undergraduate degree from Ohio Northern University in 1966. He was enrolled in the APA approved Doctoral Program in Clinical Psychology at Kent State University from 1966 – 1970. Upon graduation from Kent State, Dr. Tryon joined the Psychology Department faculty at Fordham University in 1970 as an Assistant Professor. He was promoted to Associate Professor in 1977 and to Full Professor in 1983. Licensed as a psychologist in New York State in 1973, he joined the National Register of Health Service Providers in Psychology in 1976, became a Diplomate in Clinical Psychology from the American Board of Professional Psychology (ABPP) in 1984, was promoted to Fellow of Division 12 (Clinical) of the American Psychological Association in 1994 and a fellow of the American Association of Applied and Preventive Psychology in 1996. Also in 1996 he became a Founder of the Assembly of Behavior Analysis and Therapy.
In 2003 he joined The Academy of Clinical Psychology. He was Director of Clinical Psychology Training from 1997 to 2003, and presently is in the third and final year of phased retirement. He will become Emeritus Professor of Psychology in May 2015 after 45 years of service to Fordham University. Dr. Tryon has published 179 titles, including 3 books, 22 chapters, and 140 articles in peer reviewed journals covering statistics, neuropsychology, and clinical psychology. He has reviewed manuscripts for 45 journals and book publishers and has authored 145 papers/posters that were presented at major scientific meetings. Dr. Tryon has mentored 87 doctoral dissertations to completion. This is a record number of completed dissertations at the Fordham University Graduate School of Arts and Sciences and likely elsewhere.
His academic lineage is as follows. His mentor was V. Edwin Bixenstein who studied with O. Hobart Mowrer at the University of Illinois who studied with Knight Dunlap at Johns Hopkins University who studied with Hugo Munsterberg at Harvard University who studied with Wilhelm Wundt at the University of Leipzig.
Cognitive Neuroscience and Psychotherapy: Network Principles for a Unified Theory is Dr. Tryon’s capstone publication. It is the product of more than a quarter of a century of scholarship. Additional material added after this book was printed is available at www.fordham.edu/psychology/tryon. This includes chapter supplements, a color version of Figure 5.6, and a thirteenth “Final Evaluation” chapter. He is on LinkedIn and Facebook. His email address is email@example.com.
 For those readers who wish to learn more about this technology, I recommend that they begin with a book entitled Computational Explorations in Cognitive Neuroscience: Understanding the Mind by Simulating the Brain by O’Reilly and Munakata (2000).
Dhindsa, R. S., & Goldstein, D. B. (2016). From genetics to physiology at last. Nature. doi:10.1038/nature16874
Gazzaniga, M. S. (2011). Who’s in charge? Free will and the science of the brain. New York: HarperCollins.
Hoffman, R. E., & Dobscha, S. K. (1989). Cortical pruning and the development of schizophrenia: A computer model. Schizophrenia Bulletin, 15, 477-490. doi 10.1093/schbul/15.3.477
Hoffman, R. E., & McGlashan, T. H. (1993). Parallel distributed processing and the emergence of schizophrenic symptoms. Schizophrenia Bulletin, 19, 119-139. doi 10.1093/schbul/19.1.119
Hoffman, R. E., & McGlashan, T. H. (1997). Synaptic elimination, neurodevelopment, and the mechanism of hallucinated “voices” in schizophrenia. American Journal of Psychiatry, 154, 1683-1689. doi 10.1176/ajp.154.12.1683
Insel, T. R. (2010). Rethinking schizophrenia. Nature, 187, November 11, 187-193. doi:10.1038/nature09552
Li, S.-C, Lindenberger, U., & Sikström, S. (2001). Aging cognition: From neuromodulation to representation. Trends in Cognitive Sciences, 5, 479-486. doi:10.1016/S1364-6613(00)01769-1
Li, S.-C., Naveh-Benjamin, M., & Lindenberger (2005). Aging neuromodulation impairs associative binding: A neurocomputational account. Psychological Science, 16, 445-450. doi: 10.1111/j.0956-7976.2005.01555.x
Li, S.-C, & Sikström, S. (2002). Integrative neurocomputational perspective on cognitive aging, neuromodulation, and representation. Neuroscience and Biobehavioral Reviews, 26, 795-808. doi 10.1016/S0149-7634(02)00066-0
Meyer-Lindenberg, A. (2010). From maps to mechanisms through neuroimaging of schizophrenia. Nature, 468, November 11, 194-202. doi:10.1038/nature09569
Monroe, B. M., & Read, S. J. (2008). A general connectionist model of attitude structure and change: The ACS (attitudes as constraint satisfaction) model. Psychological Review, 115, 733–759. doi: 10.1037/0033-295X.115.3.733
Neill, J. (1990). Whatever became of the schizophrenogenic mother? American Journal of Psychotherapy, 44 (4), 499-505. PMID: 2285075
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231–259. doi 10.1037/0033-295X.84.3.231
O’Reilly, R. C., & Munakata, Y. (2000). Computational explorations in cognitive neuroscience: Understanding the mind by simulating the brain. Cambridge, MA: The MIT Press.
Read, S. J., & Miller, L. C. (2002). Virtual personalities: A neural network model of personality. Personality and Social Psychology Review, 6, 357–369. doi 10.1207/S15327957PSPR0604_10
Read, S. J., Monroe, B. M., Brownstein, A. L., Yang, Y., Chopra, G., & Miller, L. C. (2010). A neural network model of the structure and dynamics of human personality. Psychological Review, 117, 61–92. doi 10.1037/a0018131.
Skar, A., Bialas, A. R., de Rivera, H., Davis, A., Hammond, T. R., Kamitaki, N., Tooley, K., Presumey, J., Baum, M., Van Doren, V., Genovese, G., Rose, S. A., Handsaker, R. E. Schizophrenia Working Group of the Psychiatric Genomics Consortium, Daly, M. J., Carroll, M. C., Stevens, B., & McCarroll, S. A. (2016). Schizophrenia risk from complex variation of complement component. Nature, doi: 10.1038/nature16549
Tryon, W. W. (2014). Cognitive neuroscience and psychotherapy: Network Principles for a Unified Theory. New York: Academic Press. http://store.elsevier.com/9780124200715
Researchers and clinicians in psychology work across a vast array of sub-disciplines, including applied psychology, addictions, cognitive psychology, developmental and educational psychology, experimental physiological psychology, forensic psychology, neuropsychology, and behavioral and cognitive therapy. For these professionals, and students as well, cross-disciplinary study is a given. For more than 75 years, Elsevier has cultivated portfolios of psychology books, eBooks, and journals covering current and critical issues in all of these areas. This vital content provides a sound basis of understanding for all those involved in this multi-faceted field.