Elsevier

Neuropsychologia

Volume 48, Issue 11, September 2010, Pages 3193-3197
Neuropsychologia

Increasing dopamine levels in the brain improves feedback-based procedural learning in healthy participants: An artificial-grammar-learning experiment

https://doi.org/10.1016/j.neuropsychologia.2010.06.024Get rights and content

Abstract

Recently, an increasing number of studies have suggested a role for the basal ganglia and related dopamine inputs in procedural learning, specifically when learning occurs through trial-by-trial feedback (Shohamy, Myers, Kalanithi, & Gluck. (2008). Basal ganglia and dopamine contributions to probabilistic category learning. Neuroscience and Biobehavioral Reviews, 32, 219–236). A necessary relationship has however only been demonstrated in patient studies. In the present study, we show for the first time that increasing dopamine levels in the brain improves the gradual acquisition of complex information in healthy participants. We implemented two artificial-grammar-learning tasks, one with and one without performance feedback. Learning was improved after levodopa intake for the feedback-based learning task only, suggesting that dopamine plays a specific role in trial-by-trial feedback-based learning. This provides promising directions for future studies on dopaminergic modulation of cognitive functioning.

Introduction

In procedural learning, acquisition of knowledge occurs gradually, and is based on an ongoing presentation of multiple stimuli and responses (thus, involving trial-by-trial feedback), exemplified by the acquisition of motor skills or language rules (Ullman, 2004). Recently, procedural learning has attracted a substantial amount of interest, emerging from neuro-imaging, animal and lesion studies, with the consensus that the basal-ganglia system and the prefrontal regions to which it projects, subserve as a neural correlate for this type of learning (Shohamy et al., 2008, Ullman, 2004). Whereas only few studies have tested healthy participants’ striatal involvement in trial-by-trial feedback learning, several patient studies have shown that basal-ganglia disorders particularly disrupted feedback-based procedural learning, as assessed by the Weather Prediction Task, where participants had to learn probabilistic cue-outcome associations over many trials (Knowlton et al., 1996a, Knowlton et al., 1996b). This finding has been corroborated by an fMRI study with healthy participants (Poldrack et al., 2001), showing engagement of the basal ganglia in a trial-by-trial feedback-based learning task. Interestingly, this activation was decreased when the same task was learned without feedback, although performance levels were similar. Since fMRI methods cannot demonstrate the necessity of the striatum in feedback-based learning, Shohamy, Myers, Onlaor, and Gluck (2004) replicated this experiment in Parkinson's disease (PD) patients, typically suffering from a loss of nigro-striatal dopamine neurons that disrupts striatal functioning (Agid, Javoy-Agid, & Ruberg, 1987). As expected, they were only impaired on the feedback-based task, but not on the non-feedback version of the same task. Another way of assessing procedural learning is through the artificial-grammar-learning (AGL) task, where participants learn a complex grammar after being exposed to positive exemplars only, hence, no feedback is involved. Here, PD patients were not impaired (Reber & Squire, 1999). Smith and McDowall (2006) manipulated the AGL task such that participants had to learn through trial-by-trial feedback, and found that PD patients are selectively impaired in a feedback-based version of the AGL task (Smith & McDowall, 2006). A consistent result throughout the literature is thus, that PD patients are impaired when learning occurs through feedback, but not when learning occurs through merely observing exemplars of a specific category (i.e., observational learning).

In the present study, we wanted to show a direct relationship between the dopamine system and feedback-based procedural learning in healthy adults, which, to our knowledge, has not been demonstrated before. To do so, we implemented two complex artificial-grammar-learning experiments, one with and one without feedback, and administered levodopa. Note that we did not intend to draw any conclusions about the role of the striatum in procedural learning. Rather, we hypothesized that increasing dopamine levels in the brain would affect learning success in the feedback-based version, but not in the observational learning version.

The outputs from the basal ganglia to the cortex are affected by dopaminergic projections from the substantia nigra to the neostriatum. Across species, the most common finding is that the dopaminergic neurons in the substantia nigra play a major role in the reward-based learning functions of the basal ganglia (Doya, 2000, Packard and Knowlton, 2002). In the current study, we set out to modulate the outputs from the basal ganglia to the frontal cortex, and hence its functioning, by increasing dopamine levels in the brain. We tested healthy participants who were administered levodopa, a precursor of dopamine, or a placebo substance. Levodopa has been demonstrated before to enhance cognitive functioning and motor learning in healthy participants (Floel et al., 2008, Knecht et al., 2004). Phasic dopamine is crucially involved in learning success (Fiorillo, 2004), whereas dopamine agonists, which affect tonic dopamine levels, lead to learning impairment (Breitenstein et al., 2006). Understanding how levodopa affects learning and memory is therefore not only of great interest from a clinical perspective, but also for gaining more insight into the neural correlates of the role of dopamine in learning and memory (Shohamy et al., 2006, Shohamy et al., 2008).

We implemented two complex artificial-grammar-learning experiments. In the first, participants had to merely observe a large amount of stimuli, all conforming to an underlying structure, i.e., grammar. In a subsequent classification test, they then had to decide whether or not novel stimuli conformed to the grammar. In the second task, participants were given performance feedback after each trial, such that they were only able to learn the grammar through the delivered feedback.

Section snippets

Participants and procedure

In a randomized, double-blind, placebo-controlled between-group study with two groups of twenty participants, we investigated the influence of taking 100 mg levodopa in combination with 25 mg of the decarboxylase inhibitor carbidopa. The placebo group received a standard placebo substance of 99.5% mannitol and .5% erosil. All medication was produced in identical capsules. Medication was given 60 min prior to participation in the experiment, to achieve optimal blood plasma levels. The timeline of

Results

From the 40 participants, we excluded two who performed below chance level (<50%), as well as two weight-matched controls, such that two equally sized groups of 18 participants each remained.

Main findings

The Levodopa group showed significantly improved learning compared to the Placebo group, but only for the feedback-based artificial-grammar-learning task. This improvement depends on the medication and is mainly driven by a significantly better ability to correctly identify grammatical items later in learning, as compared to the Placebo group.

Trial-by-trial feedback learning and the dopamine system

Feedback modulates dopamine release in the basal ganglia. A healthy range of dopamine bursts during feedback may lead to the acquisition of

Conclusion

We found improved learning for the Levodopa group as compared to the control group, but only for the artificial-grammar-learning task where learning occurred through trial-by-trial feedback. This is consistent with evidence on procedural learning tasks that trial-by-trial feedback learning is dependent on the basal ganglia. PD patients have been found to be substantially impaired on this kind of learning. Our experiments on healthy subjects support and complement the previous findings that

Acknowledgements

This research was funded by the BMBF (Federal Ministery of Education and Research, Germany), and the EU 6th Framework Program Marie Curie Research and Training Network: Language and the Brain, awarded to Meinou de Vries and Stefan Knecht. We thank Elina Sakellaridou and Julia Reinholz for their valuable help with the experiments, Caterina Breitenstein, Bianca Drager and Sandra Kamping for their contributions to project preparations, and two anonymous reviewers for their insightful comments.

References (31)

  • K. Witt et al.

    Intact artificial grammar learning in patients with cerebellar degeneration and advanced Parkinson's disease

    Neuropsychologia

    (2002)
  • Y. Agid et al.

    Biochemistry of neurotransmitters in Parkinson's disease

    Movement Disorders

    (1987)
  • C. Breitenstein et al.

    Tonic dopaminergic stimulation impairs associative learning in healthy subjects

    Neuropsychopharmacology

    (2006)
  • R. Cools et al.

    Enhanced or impaired cognitive function in Parkinson's disease as a function of dopaminergic medication and task demands

    Cerebral Cortex

    (2001)
  • S. Dethy et al.

    Microdialysis-HPLC for plasma levodopa and metabolites monitoring in parkinsonian patients

    Clinical Chemistry

    (1997)
  • Cited by (36)

    • Effects of levodopa on stimulus-response learning versus response selection in healthy young adults

      2017, Behavioural Brain Research
      Citation Excerpt :

      Bolstering the dopamine overdose hypothesis are findings of impaired learning in healthy individuals following administration of levodopa or dopamine agonists [29–37], which entirely parallel investigations in PD. At odds with this notion and in the minority in this literature, however, are reports of dopamine-related improvements in learning [38–42]. We speculate that this discrepancy in the literature is explained by the fact that typical learning paradigms often confound learning and response-selection decisions, which are differentially affected by dopaminergic therapy.

    • Musical Training as a Framework for Brain Plasticity: Behavior, Function, and Structure

      2012, Neuron
      Citation Excerpt :

      In humans, it is known that Levodopa administration can increase plasticity in the motor cortex (Kuo et al., 2008), while conversely plasticity in motor cortex is diminished in Parkinson's patients (Ueki et al., 2006). Behavioral studies have also shown that Levodopa can modulate both motor learning (Flöel et al., 2005, 2008; Rösser et al., 2008) and acquisition of an artificial language (de Vries et al., 2010). In a music training context, the produced sounds would provide direct feedback about accuracy of performance, which might be in part mediated through dopaminergic signals.

    • Linking neurogenetics and individual differences in language learning: The dopamine hypothesis

      2012, Cortex
      Citation Excerpt :

      Knowlton et al. (1996) found that subjects with basal ganglia lesions have difficulty with such tasks. Another example of a procedural learning task is the learning of an artificial grammar. de Vries et al. (2010) found in healthy adults that performance in artificial grammar learning was improved after levodopa (a precursor of DA) intake.

    • Atypical Reinforcement Learning in Developmental Dyslexia

      2022, Journal of the International Neuropsychological Society
    View all citing articles on Scopus
    View full text