ISSN 0253-2778

CN 34-1054/N

Open AccessOpen Access JUSTC Life Sciences 17 January 2024

Structural knowledge error, rather than reward insensitivity, explains the reduced metacontrol in aging

Cite this:
https://doi.org/10.52396/JUSTC-2023-0132
More Information
  • Author Bio:

    Zhaoyu Zuo received his master’s degree in Pattern Recognition and Intelligent Systems from the University of Science and Technology of China. His research mainly focuses on the competitive cooperative relationship between working memory and reinforcement learning

    Lizhuang Yang received his Ph.D. degree in Psychology from the Chinese University of Hong Kong. He is currently an Associate Professor at the Hefei Institutes of Physical Science, Chinese Academy of Sciences. His research interests include decision-making, metacognition, and individual difference

    Hai Li received his Ph.D. degree in Automation from the Northwestern Polytechnical University. He is currently a Professor at the Hefei Institutes of Physical Science, Chinese Academy of Sciences. His research focuses on neuroimaging, cognitive science, and translational technologies

  • Corresponding author: E-mail: lzyang@cmpt.ac.cn; E-mail: hli@cmpt.ac.cn
  • Received Date: 06 September 2023
  • Accepted Date: 08 October 2023
  • Available Online: 17 January 2024
  • Humans flexibly adjust their reliance on model-free (habitual) and model-based (goal-directed) strategies according to cost‒benefit trade-offs, the ability of which is known as metacontrol. Recent studies have suggested that older adults show reduced flexibility in metacontrol. However, whether the metacontrol deficit in aging is due to cognitive or motivational factors remains ambiguous. The present study investigated this issue using pupillometry recording and a sequential decision-making task with varied task structures and reward stakes. Our results revealed that older adults performed less model-based control and less flexibility when the reward stake level changed, consistent with previous studies. However, pupillometry analysis indicated that older adults showed comparable sensitivity to the reward stake. Older adults varied in task structure knowledge according to their oral reports, and the subgroup with good structural knowledge exerted a similar pattern to younger adults. Computational simulation verified that poor structure knowledge representation impaired metacontrol. These results suggest that the inflexible metacontrol in the elderly population might not be due to motivational factors but rather poor structure knowledge.
    The aging of metacontrol is due to structural knowledge errors.
    Humans flexibly adjust their reliance on model-free (habitual) and model-based (goal-directed) strategies according to cost‒benefit trade-offs, the ability of which is known as metacontrol. Recent studies have suggested that older adults show reduced flexibility in metacontrol. However, whether the metacontrol deficit in aging is due to cognitive or motivational factors remains ambiguous. The present study investigated this issue using pupillometry recording and a sequential decision-making task with varied task structures and reward stakes. Our results revealed that older adults performed less model-based control and less flexibility when the reward stake level changed, consistent with previous studies. However, pupillometry analysis indicated that older adults showed comparable sensitivity to the reward stake. Older adults varied in task structure knowledge according to their oral reports, and the subgroup with good structural knowledge exerted a similar pattern to younger adults. Computational simulation verified that poor structure knowledge representation impaired metacontrol. These results suggest that the inflexible metacontrol in the elderly population might not be due to motivational factors but rather poor structure knowledge.
    • Pupillometry analysis proves that older adults are sensitive to the level of reward.
    • Older adults with intact structural knowledge show a comparable metacontrol efficiency to younger adults.
    • Computer simulations suggest that the aging of metacontrol is due to structural knowledge errors.

  • loading
  • [1]
    Collins A G E, Cockburn J. Beyond dichotomies in reinforcement learning. Nature Reviews Neuroscience, 2020, 21 (10): 576–586. doi: 10.1038/s41583-020-0355-6
    [2]
    Kool W, Gershman S J, Cushman F A. Planning complexity registers as a cost in metacontrol. Journal of Cognitive Neuroscience, 2018, 30 (10): 1391–1404. doi: 10.1162/jocn_a_01263
    [3]
    Gilovich T, Griffin D, Kahneman D. Heuristics and Biases: The Psychology of Intuitive Judgment. Cambridge: Cambridge University Press, 2002 .
    [4]
    Kool W, Gershman S J, Cushman F A. Cost-benefit arbitration between multiple reinforcement-learning systems. Psychological Science, 2017, 28 (9): 1321–1333. doi: 10.1177/0956797617708288
    [5]
    Kool W, Cushman F A, Gershman S J. Competition and cooperation between multiple reinforcement learning systems. In: Morris R, Bornstein A, Shenhav A, editors. Goal-Directed Decision Making. New York: Academic Press, 2018 : 153–178.
    [6]
    Bolenz F, Kool W, Reiter A M, et al. Metacontrol of decision-making strategies in human aging. eLife, 2019, 8: e49154. doi: 10.7554/eLife.49154
    [7]
    Gläscher J, Daw N, Dayan P, et al. States versus rewards: dissociable neural prediction error signals underlying model-based and model-free reinforcement learning. Neuron, 2010, 66 (4): 585–595. doi: 10.1016/j.neuron.2010.04.016
    [8]
    Kool W, Botvinick M. Mental labour. Nature Human Behaviour, 2018, 2 (12): 899–908. doi: 10.1038/s41562-018-0401-9
    [9]
    Smid C R, Ganesan K, Thompson A, et al. Neurocognitive basis of model-based decision making and its metacontrol in childhood. Developmental Cognitive Neuroscience, 2023, 62: 101269. doi: 10.1016/j.dcn.2023.101269
    [10]
    Hämmerer D, Schwartenbeck P, Gallagher M, et al. Older adults fail to form stable task representations during model-based reversal inference. Neurobiology of Aging, 2019, 74: 90–100. doi: 10.1016/j.neurobiolaging.2018.10.009
    [11]
    Eppinger B, Heekeren H R, Li S C. Age-related prefrontal impairments implicate deficient prediction of future reward in older adults. Neurobiology of Aging, 2015, 36 (8): 2380–2390. doi: 10.1016/j.neurobiolaging.2015.04.010
    [12]
    Ruel A, Bolenz F, Li S C, et al. Neural evidence for age-related deficits in the representation of state spaces. Cerebral Cortex, 2023, 33 (5): 1768–1781. doi: 10.1093/cercor/bhac171
    [13]
    Vink M, Kleerekooper I, van den Wildenberg W P M, et al. Impact of aging on frontostriatal reward processing. Human Brain Mapping, 2015, 36 (6): 2305–2317. doi: 10.1002/hbm.22771
    [14]
    Spaniol J, Bowen H J, Wegier P, et al. Neural responses to monetary incentives in younger and older adults. Brain Research, 2015, 1612: 70–82. doi: 10.1016/j.brainres.2014.09.063
    [15]
    Hird E J, Beierholm U, De Boer L, et al. Dopamine and reward-related vigor in younger and older adults. Neurobiology of Aging, 2022, 118: 34–43. doi: 10.1016/j.neurobiolaging.2022.06.003
    [16]
    da Silva Castanheira K, LoParco S, Otto A R. Task-evoked pupillary responses track effort exertion: Evidence from task-switching. Cognitive, Affective, & Behavioral Neuroscience, 2021, 21 (3): 592–606. doi: 10.3758/s13415-020-00843-z
    [17]
    Rondeel E W M, van Steenbergen H, Holland R W, et al. A closer look at cognitive control: differences in resource allocation during updating, inhibition and switching as revealed by pupillometry. Frontiers in Human Neuroscience, 2015, 9: 494. doi: 10.3389/fnhum.2015.00494
    [18]
    Feher da Silva C, Hare T A. Humans primarily use model-based inference in the two-stage task. Nature Human Behaviour, 2020, 4 (10): 1053–1066. doi: 10.1038/s41562-020-0905-y
    [19]
    Zandi B, Lode M, Herzog A, et al. PupilEXT: flexible open-source platform for high-resolution pupillometry in vision research. Frontiers in Neuroscience, 2021, 15: 676220. doi: 10.3389/fnins.2021.676220
    [20]
    Santini T, Fuhl W, Kasneci E. PuRe: Robust pupil detection for real-time pervasive eye tracking. Computer Vision and Image Understanding, 2018, 170: 40–50. doi: 10.1016/j.cviu.2018.02.002
    [21]
    Kool W, Cushman F A, Gershman S J. When does model-based control pay off. PLoS Computational Biology, 2016, 12 (8): e1005090. doi: 10.1371/journal.pcbi.1005090
    [22]
    de Leeuw J R. jsPsych: A JavaScript library for creating behavioral experiments in a Web browser. Behavior Research Methods, 2015, 47: 1–12. doi: 10.3758/s13428-014-0458-y
    [23]
    Daw N D, Gershman S J, Seymour B, et al. Model-based influences on humans’ choices and striatal prediction errors. Neuron, 2011, 69 (6): 1204–1215. doi: 10.1016/j.neuron.2011.02.027
    [24]
    Rummery G A, Niranjan M. On-line Q-learning using connectionist systems. Cambridge: University of Cambridge, 1994 .
    [25]
    Bürkner P C. brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 2017, 80 (1): 1–28. doi: 10.18637/jss.v080.i01
    [26]
    Shenhav A, Botvinick M M, Cohen J D. The expected value of control: an integrative theory of anterior cingulate cortex function. Neuron, 2013, 79 (2): 217–240. doi: 10.1016/j.neuron.2013.07.007
    [27]
    Bolenz F, Profitt M F, Stechbarth F, et al. Need for cognition does not account for individual differences in metacontrol of decision making. Scientific Reports, 2022, 12 (1): 8240. doi: 10.1038/s41598-022-12341-y
    [28]
    Castro-Rodrigues P, Akam T, Snorasson I, et al. Explicit knowledge of task structure is a primary determinant of human model-based action. Nature Human Behaviour, 2022, 6 (8): 1126–1141. doi: 10.1038/s41562-022-01346-2
    [29]
    Eppinger B, Walter M, Heekeren H R, et al. Of goals and habits: age-related and individual differences in goal-directed decision-making. Frontiers in Neuroscience, 2013, 7: 253. doi: 10.3389/fnins.2013.00253
    [30]
    Walsh M M, Anderson J R. Navigating complex decision spaces: Problems and paradigms in sequential choice. Psychological Bulletin, 2014, 140 (2): 466–486. doi: 10.1037/a0033455
    [31]
    Jablonska K, Stanczyk M, Piotrowska M, et al. Age as a moderator of the relationship between planning and temporal information processing. Scientific Reports, 2022, 12 (1): 1548. doi: 10.1038/s41598-022-05316-6
    [32]
    de Wit S, van de Vijver I, Ridderinkhof K R. Impaired acquisition of goal-directed action in healthy aging. Cognitive, Affective, & Behavioral Neuroscience, 2014, 14 (2): 647–658. doi: 10.3758/s13415-014-0288-5
    [33]
    Patzelt E H, Kool W, Millner A J, et al. Incentives boost model-based control across a range of severity on several psychiatric constructs. Biological Psychiatry, 2019, 85 (5): 425–433. doi: 10.1016/j.biopsych.2018.06.018
  • 加载中

Catalog

    Figure  1.  Behavioral task. (a) State transition structure of the task. Each trial starts with a random first-stage state. Given the transition structure, each first-stage choice leads deterministically to the second-stage state. Each second-stage choice is associated with a drifting scalar reward. (b) The stake manipulation (top). One of the high-stake or low-stake tips is randomly presented at the beginning of the trial, which means that the actual benefit of the trial is several times the score shown in the feedback. Transition manipulation (bottom). The task transition did not change in stable-transition blocks, and the task transition structure changed irregularly in variable-transition blocks.

    Figure  2.  The metacontrol effect. Logistic regression weights show the influence of stakes, transitions and their interaction effect on the model-based weights in the two groups. The vertical line represents the 95% confidence interval, and the dots represent the mean.

    Figure  3.  Time series depicting the average pupil diameter of low-stake and high-stake trials (a, d) over the course of trials. Solid lines indicate the mean pupil diameter (baseline-corrected). Shaded areas indicate standard errors (SEs) of pupil diameter (baseline-corrected). Red lines on the top indicate time points of a reliable [p <0.05] positive effect for younger and older adults. Reaction time in stage 1 (b, e) and stage 2 (c, f). The error bar represents the standard error of the mean. * indicates p <0.05, ** indicates p <0.01, and *** indicates p <0.001.

    Figure  4.  Time series depicting the average pupil diameter of high (7–9), medium (3–6) and low (0–2) reward trials in the feedback phase. Solid lines indicate the mean pupil diameter (baseline-corrected). Shaded areas indicate SEs of pupil diameter (baseline-corrected).

    Figure  5.  Time series depicting the average pupil diameter of stable-transition and variable-transition trials (a, d) over the course of trials. Solid lines indicate the mean pupil diameter. Shaded areas indicate SEs of pupil diameter (baseline-corrected). Red lines on the top indicate time points of a reliable [p < 0.05] positive effect for younger and older adults. Reaction time in stage 1 (b, e) and stage 2 (c, f). The error bar represents the standard error of the mean. *** indicates p <0.001 and NS = not significant.

    Figure  6.  The metacontrol effect after grouping. Logistic regression weights show the influence of stakes, transitions and their interaction effect on the model-based weights in older adults. Older adults were grouped according to subjective reports of different levels of structural update difficulty. The vertical line represents the 95% confidence interval, and the dots represent the mean.

    Figure  7.  Model prediction. Comparison of simulated vs. empirically observed model-based weight differences. Agents were divided into three groups based on structural noise levels. The error bar represents the standard error of the mean.

    [1]
    Collins A G E, Cockburn J. Beyond dichotomies in reinforcement learning. Nature Reviews Neuroscience, 2020, 21 (10): 576–586. doi: 10.1038/s41583-020-0355-6
    [2]
    Kool W, Gershman S J, Cushman F A. Planning complexity registers as a cost in metacontrol. Journal of Cognitive Neuroscience, 2018, 30 (10): 1391–1404. doi: 10.1162/jocn_a_01263
    [3]
    Gilovich T, Griffin D, Kahneman D. Heuristics and Biases: The Psychology of Intuitive Judgment. Cambridge: Cambridge University Press, 2002 .
    [4]
    Kool W, Gershman S J, Cushman F A. Cost-benefit arbitration between multiple reinforcement-learning systems. Psychological Science, 2017, 28 (9): 1321–1333. doi: 10.1177/0956797617708288
    [5]
    Kool W, Cushman F A, Gershman S J. Competition and cooperation between multiple reinforcement learning systems. In: Morris R, Bornstein A, Shenhav A, editors. Goal-Directed Decision Making. New York: Academic Press, 2018 : 153–178.
    [6]
    Bolenz F, Kool W, Reiter A M, et al. Metacontrol of decision-making strategies in human aging. eLife, 2019, 8: e49154. doi: 10.7554/eLife.49154
    [7]
    Gläscher J, Daw N, Dayan P, et al. States versus rewards: dissociable neural prediction error signals underlying model-based and model-free reinforcement learning. Neuron, 2010, 66 (4): 585–595. doi: 10.1016/j.neuron.2010.04.016
    [8]
    Kool W, Botvinick M. Mental labour. Nature Human Behaviour, 2018, 2 (12): 899–908. doi: 10.1038/s41562-018-0401-9
    [9]
    Smid C R, Ganesan K, Thompson A, et al. Neurocognitive basis of model-based decision making and its metacontrol in childhood. Developmental Cognitive Neuroscience, 2023, 62: 101269. doi: 10.1016/j.dcn.2023.101269
    [10]
    Hämmerer D, Schwartenbeck P, Gallagher M, et al. Older adults fail to form stable task representations during model-based reversal inference. Neurobiology of Aging, 2019, 74: 90–100. doi: 10.1016/j.neurobiolaging.2018.10.009
    [11]
    Eppinger B, Heekeren H R, Li S C. Age-related prefrontal impairments implicate deficient prediction of future reward in older adults. Neurobiology of Aging, 2015, 36 (8): 2380–2390. doi: 10.1016/j.neurobiolaging.2015.04.010
    [12]
    Ruel A, Bolenz F, Li S C, et al. Neural evidence for age-related deficits in the representation of state spaces. Cerebral Cortex, 2023, 33 (5): 1768–1781. doi: 10.1093/cercor/bhac171
    [13]
    Vink M, Kleerekooper I, van den Wildenberg W P M, et al. Impact of aging on frontostriatal reward processing. Human Brain Mapping, 2015, 36 (6): 2305–2317. doi: 10.1002/hbm.22771
    [14]
    Spaniol J, Bowen H J, Wegier P, et al. Neural responses to monetary incentives in younger and older adults. Brain Research, 2015, 1612: 70–82. doi: 10.1016/j.brainres.2014.09.063
    [15]
    Hird E J, Beierholm U, De Boer L, et al. Dopamine and reward-related vigor in younger and older adults. Neurobiology of Aging, 2022, 118: 34–43. doi: 10.1016/j.neurobiolaging.2022.06.003
    [16]
    da Silva Castanheira K, LoParco S, Otto A R. Task-evoked pupillary responses track effort exertion: Evidence from task-switching. Cognitive, Affective, & Behavioral Neuroscience, 2021, 21 (3): 592–606. doi: 10.3758/s13415-020-00843-z
    [17]
    Rondeel E W M, van Steenbergen H, Holland R W, et al. A closer look at cognitive control: differences in resource allocation during updating, inhibition and switching as revealed by pupillometry. Frontiers in Human Neuroscience, 2015, 9: 494. doi: 10.3389/fnhum.2015.00494
    [18]
    Feher da Silva C, Hare T A. Humans primarily use model-based inference in the two-stage task. Nature Human Behaviour, 2020, 4 (10): 1053–1066. doi: 10.1038/s41562-020-0905-y
    [19]
    Zandi B, Lode M, Herzog A, et al. PupilEXT: flexible open-source platform for high-resolution pupillometry in vision research. Frontiers in Neuroscience, 2021, 15: 676220. doi: 10.3389/fnins.2021.676220
    [20]
    Santini T, Fuhl W, Kasneci E. PuRe: Robust pupil detection for real-time pervasive eye tracking. Computer Vision and Image Understanding, 2018, 170: 40–50. doi: 10.1016/j.cviu.2018.02.002
    [21]
    Kool W, Cushman F A, Gershman S J. When does model-based control pay off. PLoS Computational Biology, 2016, 12 (8): e1005090. doi: 10.1371/journal.pcbi.1005090
    [22]
    de Leeuw J R. jsPsych: A JavaScript library for creating behavioral experiments in a Web browser. Behavior Research Methods, 2015, 47: 1–12. doi: 10.3758/s13428-014-0458-y
    [23]
    Daw N D, Gershman S J, Seymour B, et al. Model-based influences on humans’ choices and striatal prediction errors. Neuron, 2011, 69 (6): 1204–1215. doi: 10.1016/j.neuron.2011.02.027
    [24]
    Rummery G A, Niranjan M. On-line Q-learning using connectionist systems. Cambridge: University of Cambridge, 1994 .
    [25]
    Bürkner P C. brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 2017, 80 (1): 1–28. doi: 10.18637/jss.v080.i01
    [26]
    Shenhav A, Botvinick M M, Cohen J D. The expected value of control: an integrative theory of anterior cingulate cortex function. Neuron, 2013, 79 (2): 217–240. doi: 10.1016/j.neuron.2013.07.007
    [27]
    Bolenz F, Profitt M F, Stechbarth F, et al. Need for cognition does not account for individual differences in metacontrol of decision making. Scientific Reports, 2022, 12 (1): 8240. doi: 10.1038/s41598-022-12341-y
    [28]
    Castro-Rodrigues P, Akam T, Snorasson I, et al. Explicit knowledge of task structure is a primary determinant of human model-based action. Nature Human Behaviour, 2022, 6 (8): 1126–1141. doi: 10.1038/s41562-022-01346-2
    [29]
    Eppinger B, Walter M, Heekeren H R, et al. Of goals and habits: age-related and individual differences in goal-directed decision-making. Frontiers in Neuroscience, 2013, 7: 253. doi: 10.3389/fnins.2013.00253
    [30]
    Walsh M M, Anderson J R. Navigating complex decision spaces: Problems and paradigms in sequential choice. Psychological Bulletin, 2014, 140 (2): 466–486. doi: 10.1037/a0033455
    [31]
    Jablonska K, Stanczyk M, Piotrowska M, et al. Age as a moderator of the relationship between planning and temporal information processing. Scientific Reports, 2022, 12 (1): 1548. doi: 10.1038/s41598-022-05316-6
    [32]
    de Wit S, van de Vijver I, Ridderinkhof K R. Impaired acquisition of goal-directed action in healthy aging. Cognitive, Affective, & Behavioral Neuroscience, 2014, 14 (2): 647–658. doi: 10.3758/s13415-014-0288-5
    [33]
    Patzelt E H, Kool W, Millner A J, et al. Incentives boost model-based control across a range of severity on several psychiatric constructs. Biological Psychiatry, 2019, 85 (5): 425–433. doi: 10.1016/j.biopsych.2018.06.018

    Article Metrics

    Article views (303) PDF downloads(864)
    Proportional views

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return