The neuroscience of habituated motivation - PowerPoint Presentation

1. The neuroscience of habituated motivationAlberto Masala (PI), SND, Univ. Paris SorbonneDaniel Andler, SND, Univ. Paris SorbonneJean Denizeau, MBB, ICM, Univ. P. & M. CurieMathias Pessiglione, MBB, ICM, Univ. P. & M. Curie

2. Two interlocked aimsto buttress the Aristotelian theory of cultivation and motivational habituation (apprenticeship) by providing a neuroscientific account of its enabling mechanisms;to contribute to the integration of moral philosophy and cognitive neuroscience in a novel way, based on a recent turn in cognitive science.

3. One questionGiven thatthe virtuous apprenticeship path is often either not taken or soon abandonedthe virtuous apprenticeship path is sometimes takenWhat are the conditions under which apprenticeship gets underway?

4. How we propose to answerRecently developed models of cognitive architecture—predictive HBMs—seem precisely poised to provide at least the beginnings of very different kind of answer, a naturalistic answer. Our team combines the necessary competencies:Alberto Masala, philosophy (virtue theory)Jean Daunizeau, theoretical neuroscience (Bayesian models)Mathias Pessiglione, biological neuroscience (motivation, advanced skills acquisition)Daniel Andler, philosophy (models in cognitive science)

5. Overcoming the fragility of virtueWe want to discover, model and test factors that would unlock our ability to cultivate complex motivational habits.

6. Virtue as Skill (MacIntyre, Annas)Basic movements & stereotypical attitudes Good technique & grasp of major priorities in a matchSuperior technique & deep understandingAwkard Interventions & basic emphatyDecent coping strategies & good emphatyResolute action & subtle moral sensitivityBeginnerIntermediateMasterTennis PlayerCompassionate Teacher

7. Fragility of motivational habituationLosing out to the forces of evil…Egoism, hedonsim, social pressure, situationist scenarios….and laziness (within non-moral mastery)Stagnation of professionals K. A. ERICSSON, The Influence of Experience and Deliberate Practice on the Development of Superior Expert Performance, 2006 Routinized reactions of experts M. BILALIC & al. “Inflexibility of experts”, 2008 Failure to transfer knowledge D. BRANSFORD, L. SCHWARTZ, “Rethinking Transfer”, 1999

8. Default conservatism: narrow, stagnating context locked skills built at minimal cost for specific goalsInvestment in complexity: subtle, flexible mastery and motivational habituation.Right Conditions

9. Motivational habituation in Learning SciencesA.Masala « Mastering Wisdom », in A. Masala & J. Webber, eds. From Personality to Virtue (OUP forthcoming)Knowledge-Building CommunitiesInterest in learning and understanding is instilled through gradual motivational habituationCarl Bereiter & Marlene Scardamalia

10. Cognitive learning sciences & psychology of expertiseOur project: computational neurosciencesMore specific definition of the right apprenticeship conditions:Improvement over common sense & phenomenologyBasic obstacles and biases that stop the apprenticeship process

11. The search for neurocognitive mechanisms & the promise of predictive Bayesianism The Bayesian tsunami in cognitive scienceCombining the best of 2 worlds:classicism’s ability to deal with complex structured representationsconnectionism’s ability to account for learningHBM: Hierarchical Bayesian ModelPredictive coding: “Let me guess and if I’m wrong I’ll make the necessary adjustments” Friston, K. (2008). Hierarchical Models in the Brain. PLoS Computational Biology, 4(11) Tenenbaum, J. B., Kemp, C., Griffiths, T. L., & Goodman, N. D. (2011). How to Grow a Mind: Statistics, Structure, and Abstraction. Science, 331(6022), 1279–1285

12. HBMs at workStarting with the highest (deepest) layer, each layer issues a prediction on the input of the next one below. When the last prediction hits the last layer, the error is ‘lazily’ retropropagated upwardThese ideas have been highly productive in the field of visual perception, and are now being extended to a wide variety of higher cognitive tasks, such as categorization, predictions about everyday events and, importantly, causal reasoning. Mathys, C., Daunizeau, J., Friston, K. J., & Stephan, K. E. (2011). A Bayesian foundation for for individual learning under uncertainty. Frontiers in Human Neuroscience, 5, 39

13. Understanding Aristotelian apprenticeship: HBMs’ advantageHBMs embody conservatism: deep learning is costly.An HBM, exposed in the right conditions to the right learning regimen, will undergo deep change.HBMs can account for inter-individual differences, as well as temporal intra-individual differences in the capacity for deep learning.HBMs seem to be able to handle in an integrated manner the motivation and the knowledge dimensions. Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36(03), 181–204

14. Applying this framework to our problemBy no means a trivial task.Establishing a conceptual common ground, between philosophy and neuroscience, from which to attack this problem, requires a considerable effort.At the same time, we want to provide an ‘existence proof’, showing on a special case that it can be done and that it is profitable.

15. Applying this framework to our problem, #1Our long-term goal: identify the subtle factors that mediate the development of sophisticated skills, in their interconnected cognitive and motivational dimensions.First step: Focus on motivation, and examine what neuroscience and Bayesian modeling can tell us about akrasia in normal subjects.Two-pronged attack: HBM modelingPsychological and neuropsychological evidence:What role do errors in expectations of effort / reward / delay play in akrasia ? Is there a correlation between types/magnitudes of errors and proneness to akrasia? What can we learn from motivation diminution disorders such as aboulia, apathy, auto-activation deficit, athymormia or apraxia ?

16. Putting this intuition to work, #2A behavioral experiment along the following lines, aiming at testing hypotheses bearing on the conditions under which an akratic bias can be overcome.

17. Evaluation taskChoice taskRedundant informationUncertaintyvolatilityChoice taskDetermination of preferencesAkratic bias (e.g. effort bias)Variation of statistical structure in Learning conditionsHas akratic bias disappeared?

18. Potential hurdlesThe matter of levels: bridging the subpersonal account of cognitive neuroscience and the personal account of virtue theory, psychology and phenomenology.The blending of learning and motivation: despite its being on the computational neuroscientist’s horizon as a theoretical possibility, it is not as yet part of the experimentalist’s mindset.

19. Help: ideasLevels: a nagging problem for the entire field of cognitive scienceyet the neurocomputational tradition, from Helmholz to contemporary frameworks, provides hints, both negative (e.g. McCulloch & Pitts’ ‘logical calculus of the ideas immanent in nervous activity’) and positive (e.g. Smolensky’s dual system in “The proper treatment of connectionism”) neuroscientists’ interest in consciousness puts the (distinct yet connected) problem on their agendaLearning/motivation: pragmatism and Friston’s “action-oriented predictive processing”:inquiry as the activity of an engaged agent facing a problem and seeking to restore a state of harmony around her.

20. Help: peoplethe thriving cogsci community in Pariswith a particularly strong interdisciplinary tradition (Institute of Cognitive Studies, Ecole normale supérieure; ICM Pitié, UPMC; etc.)a strong neurocomputational school, straddling physics and neurosciencea strong school in philosophy of mind and philosophy of cognitive science.

21. FIN