arxiv: 2604.15363 · v1 · submitted 2026-04-13 · 🧬 q-bio.NC · cs.LG

Recognition: unknown

Machine learning approaches to uncover the neural mechanisms of motivated behaviour: from ADHD to individual differences in effort and reward sensitivity

Nam Trinh

Authors on Pith no claims yet

Pith reviewed 2026-05-10 15:05 UTC · model grok-4.3

classification 🧬 q-bio.NC cs.LG

keywords ADHDmachine learningEEGdiffusion MRIstructural MRIeffort sensitivityreward sensitivityfronto-parietal circuits

0 comments

The pith

Machine learning identifies fronto-parietal circuits as central to effort and reward processing in ADHD.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This thesis applies machine learning to EEG and neuroimaging data to uncover the brain mechanisms behind motivated behaviour, focusing on ADHD and variations in how people value effort and rewards. One study shows that EEG recordings during a task classify adults with ADHD more accurately than resting-state data, with gamma-band activity in fronto-central and parietal regions as the strongest predictors. Another links white matter integrity in tracts connected to the supplementary motor area with computational measures of effort and reward sensitivity, while a third uses grey matter volumes to decode reward sensitivity and apathy levels via machine learning. If these patterns hold, they could provide neural biomarkers to refine ADHD diagnosis and guide personalised interventions for motivational issues.

Core claim

The paper claims that fronto-parietal circuits are central to effort valuation and reward processing, as demonstrated by machine learning models applied across EEG, diffusion MRI, and structural MRI in studies of ADHD and individual differences. Task-based EEG gamma power over fronto-central and parietal areas best classified ADHD versus controls. SMA-connected white matter tracts associated with effort and reward parameters, and grey matter volumes robustly predicted reward sensitivity and subclinical apathy. These results suggest the circuits may serve as biomarkers for improving diagnostic accuracy in ADHD and motivational impairments and for directing neurotechnological interventions.

What carries the argument

Machine learning models applied to task-based and resting-state EEG, diffusion MRI white matter integrity, and structural MRI grey matter volumes to classify ADHD and predict effort and reward sensitivities, converging on fronto-parietal circuits.

Load-bearing premise

The identified machine learning features and brain associations will generalise beyond the particular tasks, scanners, and participant groups studied here without significant overfitting or confounding variables.

What would settle it

An independent study with new participants, different scanners, or varied tasks fails to replicate the superior performance of task-based EEG classifiers for ADHD or the specific associations between SMA white matter tracts and effort/reward parameters.

Figures

Figures reproduced from arXiv: 2604.15363 by Nam Trinh.

**Figure 3.1.** Figure 3.1: Experimental design, EEG data acquisition, preprocessing, feature extraction and classification pipeline. A. EEG was recorded during two conditions: resting state with eyes closed and performance of the stop-signal task. B. In the stop-signal task, participants responded to directional arrows on “Go” trials and withheld responses when a stop signal appeared. C. EEG data were preprocessed with bandpass (… view at source ↗

**Figure 3.2.** Figure 3.2: Statistical analysis of classification performance with restingstate and stop-signal task EEG. A. Left, distribution of AUC across models compared with chance. Middle, permutation test with 1,000 iterations showed no significant deviation from chance (Monte Carlo p = 0.064). Right, true-label performance was significantly greater than chance (t = 124.99, p < 0.001) and higher than permuted-label perfor… view at source ↗

**Figure 3.3.** Figure 3.3: Top most important EEG features for machine learning evaluated with resting-state and stop-signal task EEG data. Permutationbased feature importance analysis revealed distinct spectral–spatial profiles across paradigms. A. In the resting-state condition, the highest-ranking predictors were theta/beta ratio (TBR) measures over parietal and central electrodes, with additional contributions from occipito… view at source ↗

**Figure 4.1.** Figure 4.1: Experimental protocol and data analyses. A. Diffusion MRI (dMRI) data acquisition. dMRI data were acquired to assess white matter microstructure, including FA and MD. B. Effort-based decision-making task and computational modelling of behaviour. Participants completed an effort-based decisionmaking task, choosing whether to perform biceps contractions of varying effort levels (20%, 40%, 60%, or 80% o… view at source ↗

**Figure 4.2.** Figure 4.2: Effort and reward sensitivity show substantial inter-individual variability [PITH_FULL_IMAGE:figures/full_fig_p086_4_2.png] view at source ↗

**Figure 4.3.** Figure 4.3: Microstructure in tracts connected to frontal valuation regions showing negative associations with effort sensitivity (βEffort). (Caption continued on next page.) Pre-examination copy - 021225 89 [PITH_FULL_IMAGE:figures/full_fig_p089_4_3.png] view at source ↗

**Figure 4.4.** Figure 4.4: Microstructure in a tract connected to frontal valuation regions showing a positive association with effort sensitivity (βEffort). (i) A significant cluster (green) exhibited a positive correlation between FA and βEffort (MNI: -9.95, 34.7, 11.22). (ii) Probabilistic tract labelling identified the anterior cingulum. To clarify which medial frontal regions were specifically connected by this cluster, we p… view at source ↗

**Figure 4.5.** Figure 4.5: Microstructure in tracts connected to frontal valuation regions and sensorimotor structures and showing negative associations with reward sensitivity (βReward) (Figure caption continued on next page.) Pre-examination copy - 021225 94 [PITH_FULL_IMAGE:figures/full_fig_p094_4_5.png] view at source ↗

**Figure 4.6.** Figure 4.6: Exclusive overlap of white matter clusters associated with effort and reward sensitivity within the SMA portion of the corticospinal tract. FA analyses identified two significant clusters within the SMA portion of the left corticospinal tract. The cluster associated with βEffort (red) was centered at MNI: X = –13.95, Y = –10.01, Z = 58.02 (94 mm³), while the cluster associated with βReward (green) was ce… view at source ↗

**Figure 4.7.** Figure 4.7: B). Finally, one other cluster was located in a tract connected to motor [PITH_FULL_IMAGE:figures/full_fig_p096_4_7.png] view at source ↗

**Figure 4.8.** Figure 4.8: White matter microstructure predicts individual differences in effort and reward sensitivity. (Caption continued next page.) Pre-examination copy - 021225 100 [PITH_FULL_IMAGE:figures/full_fig_p100_4_8.png] view at source ↗

**Figure 4.9.** Figure 4.9: Decoding effort and reward sensitivity is classifier-independent and cluster-specific. A. Classifier-independent and cluster-specific decoding of βEffort. Left panel (blue): Area under the curve (AUC) values for 12 machine learning classifiers trained on microstructural measures from the 5 clusters significantly associated with βEffort. Decoding performance was significantly above chance for nearly all … view at source ↗

**Figure 4.10.** Figure 4.10: SMA-connected clusters dominate in terms of predictive power, with distributed circuits contributing to reward sensitivity decoding. A. Cluster contributions and optimal cluster set for decoding βEffort. Left panel: Feature importance analysis, where “features” denote the individual white matter clusters used as input variables in the model (a standard machine learning term for predictors). The most pr… view at source ↗

**Figure 5.1.** Figure 5.1: GMV from structural MRI and correlational analysis method. A. Structural MRI data were acquired from 45 healthy participants and processed using FreeSurfer for volumetric segmentation and the Brainnetome Atlas for parcellation (Caption continued next page). Pre-examination copy - 021225 119 [PITH_FULL_IMAGE:figures/full_fig_p119_5_1.png] view at source ↗

**Figure 5.2.** Figure 5.2: Higher effort sensitivity was associated with greater GMV in [PITH_FULL_IMAGE:figures/full_fig_p124_5_2.png] view at source ↗

**Figure 5.3.** Figure 5.3: Reward sensitivity (βReward) showed positive correlations with GMVs in A. the right postcentral area of the superior parietal lobule (MNI coordinates: 23, −43, 67; R = 0.428, p = 0.00422) and B. the left ventromedial putamen (MNI coordinates: −23, 7, −4; R = 0.356, p = 0.0193). Individuals with lower regional volume in these areas exhibited reduced reward sensitivity. Scatter plots show partial correlati… view at source ↗

**Figure 5.4.** Figure 5.4: Apathy scores and GMV correlation (Caption continued on next page) [PITH_FULL_IMAGE:figures/full_fig_p126_5_4.png] view at source ↗

**Figure 5.5.** Figure 5.5: Classifier performance compared to chance level across classification tasks. A, B. For effort sensitivity and reward sensitivity classifications, all classifiers except the support vector machine (SVM) achieved performance significantly higher than chance. C. In the apathy level classification, all models performed significantly above chance. Boxplots show the distribution of the area under the receiv… view at source ↗

**Figure 5.6.** Figure 5.6: Permutation test on apathy score, effort sensitivity and reward sensitivity classification performance. Caption continued on next page. Pre-examination copy - 021225 129 [PITH_FULL_IMAGE:figures/full_fig_p129_5_6.png] view at source ↗

**Figure 5.7.** Figure 5.7: Feature importance analysis of the best classifiers among three classification experiments. Feature importance values were derived from the best-performing classifiers predicting individual differences in apathy, effort sensitivity, and reward sensitivity from regional GMVs. A. For apathy classification, the ventrolateral fusiform gyrus and supplementary motor area showed the strongest contributions. B.… view at source ↗

read the original abstract

Motivated behaviour relies on the brain's capacity to evaluate effort and reward. Dysregulation within these processes contributes to a spectrum of conditions, from hyperactivity in attention-deficit/hyperactivity disorder (ADHD) to diminished goal-directed behaviour in apathy. This thesis investigates the neural mechanisms underlying ADHD using electroencephalography (EEG) and examines individual differences in effort and reward sensitivity using neuroimaging, applying machine learning approaches through three main studies. In Study 1, task-based and resting-state EEG were employed with machine learning models to classify adult individuals with ADHD and healthy controls. Machine learning classifiers trained on task-based EEG during a stop signal task outperformed those trained on resting-state EEG, with the strongest predictive features arising from gamma-band spectral power over fronto-central and parietal regions. In Study 2, diffusion MRI and whole-brain permutation-based analyses identified associations between white matter integrity and computationally modelled parameters reflecting effort and reward sensitivity, with SMA-connected tracts emerging as a central hub. In Study 3, grey matter volumes from structural T1-weighted MRI were used to examine correlates of effort sensitivity, reward sensitivity, and subclinical apathy, with machine learning confirming robust decoding of reward sensitivity and apathy levels. Across studies, fronto-parietal circuits emerged as central to effort valuation and reward processing. These findings may serve as neural biomarkers for improving diagnostic accuracy in ADHD and motivational impairments, and for guiding personalised neurotechnological interventions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This thesis applies routine ML and neuroimaging pipelines to ADHD and effort-reward questions, finding consistent fronto-parietal signals but with no validation details or out-of-sample checks in the abstract.

read the letter

The main thing to know is that this thesis runs three standard analyses: ML classifiers on task and resting EEG to separate ADHD from controls, diffusion MRI linked to computational effort and reward parameters, and structural MRI with ML for reward sensitivity and apathy. Fronto-parietal features and SMA-connected tracts come up across them, which gives a coherent thread even if the methods are off-the-shelf.

Referee Report

3 major / 2 minor

Summary. This manuscript reports three studies applying machine learning to EEG and neuroimaging data to investigate neural mechanisms of motivated behavior, focusing on ADHD classification and individual differences in effort/reward sensitivity. Study 1 shows task-based EEG (stop-signal task) classifiers outperform resting-state EEG for ADHD vs. control classification, with key features in gamma-band spectral power over fronto-central and parietal regions. Study 2 uses diffusion MRI and whole-brain permutation analyses to link white-matter integrity (especially SMA-connected tracts) to computationally modeled effort and reward parameters. Study 3 applies structural MRI grey-matter volumes and ML to decode reward sensitivity and subclinical apathy. The work concludes that fronto-parietal circuits are central to effort valuation and reward processing and may serve as biomarkers for ADHD diagnosis and motivational impairments.

Significance. If the reported associations and classification advantages prove robust after proper validation, the work could contribute neural biomarkers for ADHD and apathy, bridging computational modeling of effort/reward with multi-modal imaging and ML. The multi-study design linking task-based EEG, white-matter tracts, and grey-matter correlates is a strength, as is the focus on fronto-parietal hubs. These elements could inform personalized interventions if generalizability is demonstrated. Current impact is constrained by missing validation details.

major comments (3)

[Abstract] Abstract: The claim that task-based EEG classifiers outperformed resting-state ones, with strongest features from gamma-band power over fronto-central/parietal regions, supplies no cross-validation details, sample sizes, performance metrics with effect sizes, or multiple-comparison corrections. These omissions are load-bearing for the central classification-performance claim and prevent assessment of whether results exceed chance or overfitting.
[Abstract] Abstract: The biomarker interpretation—that fronto-parietal circuits can improve diagnostic accuracy in ADHD and motivational impairments—requires evidence of generalizability beyond the three specific cohorts, tasks, and scanners. No out-of-sample testing, external validation, or confound controls (e.g., medication status, head motion, IQ) are mentioned, which is essential for the claim that identified features reflect general mechanisms rather than sample-specific effects.
[Study 2] Study 2 (permutation-based analyses): The reported associations between SMA-connected white-matter tracts and effort/reward parameters need explicit confirmation of whole-brain multiple-comparison correction and confound regression; without these, the central-hub interpretation risks being driven by unmeasured variables or uncorrected tests.

minor comments (2)

[Abstract] Abstract: Specify the exact ML algorithms (e.g., SVM, random forest) and the computational modeling framework used to derive effort/reward parameters; this would improve reproducibility and clarity.
[Discussion] The manuscript should include a dedicated limitations paragraph addressing generalizability across scanners, tasks, and populations, as well as the absence of longitudinal or interventional validation.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback on our manuscript. We have addressed each major comment below and will revise the manuscript accordingly to improve clarity, rigor, and transparency.

read point-by-point responses

Referee: [Abstract] Abstract: The claim that task-based EEG classifiers outperformed resting-state ones, with strongest features from gamma-band power over fronto-central/parietal regions, supplies no cross-validation details, sample sizes, performance metrics with effect sizes, or multiple-comparison corrections. These omissions are load-bearing for the central classification-performance claim and prevent assessment of whether results exceed chance or overfitting.

Authors: We agree that the abstract must be self-contained and include sufficient methodological and statistical details to evaluate the classification results. Although these elements (cross-validation procedure, sample sizes, performance metrics with effect sizes, and multiple-comparison corrections via permutation testing and FDR) are reported in the full text of Study 1, we will revise the abstract to concisely summarize them. This will allow readers to assess whether the task-based EEG advantage exceeds chance levels and overfitting risks. revision: yes
Referee: [Abstract] Abstract: The biomarker interpretation—that fronto-parietal circuits can improve diagnostic accuracy in ADHD and motivational impairments—requires evidence of generalizability beyond the three specific cohorts, tasks, and scanners. No out-of-sample testing, external validation, or confound controls (e.g., medication status, head motion, IQ) are mentioned, which is essential for the claim that identified features reflect general mechanisms rather than sample-specific effects.

Authors: We acknowledge that the biomarker interpretation must be presented with appropriate caution regarding generalizability. The current studies are limited to the described cohorts and acquisition protocols, and we do not provide external validation or out-of-sample testing. In the revised manuscript, we will temper the language in the abstract and discussion, explicitly note the absence of external validation, detail any confound controls already implemented (e.g., medication status, head motion, IQ where available), and add a dedicated limitations section highlighting the need for future multi-site replication to establish broader applicability. revision: partial
Referee: [Study 2] Study 2 (permutation-based analyses): The reported associations between SMA-connected white-matter tracts and effort/reward parameters need explicit confirmation of whole-brain multiple-comparison correction and confound regression; without these, the central-hub interpretation risks being driven by unmeasured variables or uncorrected tests.

Authors: We agree that explicit confirmation of the statistical controls is essential. In Study 2, whole-brain permutation testing with threshold-free cluster enhancement was used to correct for multiple comparisons across the brain, and linear regression was applied to remove variance associated with confounds including age, sex, and head motion before assessing associations with effort and reward parameters. We will revise the methods and results sections to state these procedures explicitly and confirm that the SMA-connected tract findings survived correction, thereby strengthening the central-hub interpretation. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical ML associations on independent cohorts with no derivations or self-referential predictions

full rationale

The manuscript reports three separate empirical studies applying standard ML classifiers and permutation tests to EEG, dMRI, and sMRI data. No equations, first-principles derivations, or model-based predictions appear; all reported features, accuracies, and tract associations are direct outputs of data-driven fitting on the described cohorts. No self-citation chain, ansatz smuggling, or renaming of known results is invoked to justify the central claims. The work is therefore self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The manuscript is an empirical thesis containing no mathematical derivations, new physical constants, or postulated entities; all reported quantities arise from standard statistical and machine-learning fitting procedures applied to acquired brain data.

pith-pipeline@v0.9.0 · 5553 in / 1138 out tokens · 34570 ms · 2026-05-10T15:05:52.366219+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

12 extracted references · 4 canonical work pages

[1]

Amini and S

M. Amini and S. Chang. A review of machine learning approaches for high di- mensional process monitoring. InProceedings of the 2018 Industrial and Systems Engineering Research Conference,

2018
[2]

doi: https://doi

ISSN 1388-2457. doi: https://doi. org/10.1016/S1388-2457(01)00488-6. URLhttps://www.sciencedirect.com/ science/article/pii/S1388245701004886. M. Colom, I. Kraev, A. K. Stramek, I. B. Loza, C. L. Rostron, C. J. Heath, E. J. Dommett, and B. F. Singer. Conditioning-and reward-related dendritic and presy- naptic plasticity of nucleus accumbens neurons in male...

work page doi:10.1016/s1388-2457(01)00488-6
[3]

S. B. Floresco, D. R. Montes, M. Maric, and M. van Holstein. Differential con- tributions of nucleus accumbens subregions to cue-guided risk/reward decision making and implementation of conditional rules.Journal of Neuroscience, 38(8): 1901–1914,

1901
[4]

T. Hai, R. Swansburg, C. K. Kahl, H. Frank, K. Stone, J.-F. Lemay, and F. P. MacMaster. Right superior frontal gyrus cortical thickness in pediatric adhd. Journal of Attention Disorders, 26(14):1895–1906,

1906
[5]

L. M. Jenkins, L. Wang, H. Rosen, and S. Weintraub. A transdiagnostic review of neuroimaging studies of apathy and disinhibition in dementia.Brain, 145(6): 1886–1905,

1905
[6]

Kiiski, M

H. Kiiski, M. Bennett, L. M. Rueda-Delgado, F. R. Farina, R. Knight, R. Boyle, D. Roddy, K. Grogan, J. Bramham, C. Kelly, et al. Eeg spectral power, but not theta/beta ratio, is a neuromarker for adult adhd.European Journal of Neuro- science, 51(10):2095–2109,

2095
[7]

doi: 10.1037/0894-4105.13.3.424. A. Morin, J. Bernard, G. Carle, A. Ponchel, C. Azuar, G. Uggocioni, T. Mauras, D. Bendetowicz, M. Camus, A. Delcul, et al. Chronic apathy following a major depressive episode: what is it?Cortex,

work page doi:10.1037/0894-4105.13.3.424
[8]

doi: https: //doi.org/10.1016/j.parkreldis.2008.07.006

ISSN 1353-8020. doi: https: //doi.org/10.1016/j.parkreldis.2008.07.006. URLhttps://www.sciencedirect. com/science/article/pii/S1353802008002319. F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blon- del, P. Prettenhofer, R. Weiss, V. Dubourg, et al. Scikit-learn: Machine learning in python.the Journal of machine Learning rese...

work page doi:10.1016/j.parkreldis.2008.07.006 2008
[9]

Peretzke, P

R. Peretzke, P. F. Neher, G. A. Brandt, S. Fritze, S. Volkmer, J. Daub, G. Northoff, J. Bohn, Y. Kirchhoff, S. Roy, et al. Deciphering white matter microstructural alterations in catatonia according to icd-11: replication and machine learning analysis.Molecular psychiatry, 30(5):2095–2107,

2095
[10]

Sheelakumari, C

R. Sheelakumari, C. Bineesh, T. Varghese, C. Kesavadas, J. Verghese, and P. S. Mathuranath. Neuroanatomical correlates of apathy and disinhibition in be- havioural variant frontotemporal dementia.Brain imaging and behavior, 14(5): 2004–2011,

2004
[11]

Pingouin: statistics in Python , volume =

doi: 10.21105/joss.01026. URLhttps://doi.org/10.21105/joss. 01026. M. A. Van Der Meer and A. D. Redish. Covert expectation-of-reward in rat ventral striatum at decision points.Frontiers in integrative neuroscience, 3:458,

work page doi:10.21105/joss.01026
[12]

Performance was quantified using AUC, and statistical comparisons were corrected for multiple comparisons using the false discovery rate (FDR)

Pre-examination copy - 021225 194 Appendices .1 Appendix materials for chapter 3 Ablation study method:In addition to the main analysis, we examined the impact of targeted feature removal on classification performance for resting-state and task-related EEG data (see Figure .1.1). Performance was quantified using AUC, and statistical comparisons were corre...

2011