Affective Learning & Endophenotypes

Understanding how individual differences in Pavlovian conditioning determine who is vulnerable to cue-driven overconsumption, and who responds to behavioural interventions.

The problem: why do the same cues control some people more than others?

Environmental cues associated with palatable foods or addictive substances can acquire powerful motivational pull through Pavlovian conditioning. Over time, these cues become "motivational magnets" that capture attention, trigger approach, and elicit craving. Yet individuals differ enormously in the degree to which cues come to control their behaviour. Some people walk past a bakery without a second thought; others experience an almost irresistible urge to enter. Understanding what drives this difference is essential both for basic science and for the design of effective interventions.

Our programme investigates the Pavlovian endophenotypes that underlie this variability. We study how individual differences in the way people learn from reward-predictive cues shape food reward processing, craving vulnerability, and responsiveness to cognitive training interventions.

Sign-tracking and goal-tracking: two modes of learning from reward cues

In Pavlovian conditioning, some individuals develop conditioned responses directed at the predictive cue itself (sign-trackers), while others direct responses toward the reward delivery location (goal-trackers). This distinction, originally characterised in rodent autoshaping research (Flagel et al., 2009, 2011) and now translated to humans (Schad et al., 2020; Heck et al., 2024), reflects fundamentally different learning strategies. Sign-trackers attribute incentive salience to the cue, transforming it into a "wanted" object. Goal-trackers treat the cue as a mere informational predictor.

At the neural level, sign-tracking depends on phasic dopamine signalling in the nucleus accumbens core and is associated with model-free (habitual) reinforcement learning, while goal-tracking recruits prefrontal goal-directed systems. Critically, sign-trackers show greater susceptibility to cue-triggered relapse and overconsumption (Colaizzi et al., 2020), making the sign-tracking/goal-tracking (ST/GT) continuum a candidate endophenotype for personalised intervention design.

Our contribution: linking Pavlovian phenotype to training responsiveness

A central finding from our lab is that Pavlovian learning style moderates the effectiveness of Go/NoGo response training. In our recent Registered Report (Tapparel et al., in press), we demonstrated that sign-tracking bias, measured via the Pavlovian-Instrumental Discrimination Index (PIDI), explains 21% of the variance in training-induced devaluation of food cues (p = .008). Individuals with stronger sign-tracking tendencies showed larger reductions in explicit wanting after inhibition training. This is the first direct evidence linking Pavlovian learning phenotype to response training outcomes in humans.

This finding builds on our broader programme of preregistered trials showing that Go/NoGo training to sugary-drink cues produces robust devaluation effects (d > 0.5) and measurable consumption reductions (Najberg et al., 2021, 2023). Together, these results suggest that sign-trackers, whose cue-centred learning amplifies the motivational properties of appetitive cues, are precisely the individuals for whom motor-based interventions are most effective, because the same learning bias that makes them vulnerable also makes them responsive to cue-targeted retraining.

Neural mechanisms: topographic EEG and the value-update window

A distinctive feature of our approach is the use of high-density topographic EEG with electrical neuroimaging to characterise the neural dynamics of affective learning and value updating. Rather than relying on single-electrode analyses, we employ Topographic Analysis of Variance (TANOVA) and Topographic Analysis of Covariance (TANCOVA) to capture full spatial configurations of brain activity with millisecond resolution (Koenig et al., 2011; Michel & Murray, 2012). This methodology, developed through a decade of work in our lab (Hartmann et al., 2016; De Pretto et al., 2019; Sallard et al., 2018), allows us to dissociate changes in response strength from qualitative reorganisation of the underlying neural networks.

Our data consistently show that Go/NoGo training modulates the N2 component (150-300 ms), which reflects conflict detection and resolution in frontocentral regions. Crucially, training induces both quantitative enhancement (Global Field Power) and qualitative network reorganisation (Global Map Dissimilarity) of N2 responses to trained cues (Tapparel et al., in press). Source estimation localises these effects to the insula and supplementary motor area, regions critical for interoceptive valuation and motor-reward integration. This pattern indicates genuine reassignment of affective value rather than mere enhancement of motor inhibition, and it co-occurs with reductions in explicit liking, linking neural plasticity to behavioural outcome.

Preliminary evidence further suggests that the precise timing at which an individual processes the motivational value of a cue, as captured by TANCOVA, predicts how strongly they benefit from training. This opens the possibility that individual differences in the temporal dynamics of value processing may explain a substantial portion of the variability in training responsiveness.

Current directions

Our current work extends these findings in several directions. We are investigating how individual differences along the ST/GT continuum interact with the temporal parameters of training and with the semantic content of trained actions to jointly determine intervention outcomes. We are also examining whether the ST/GT framework generalises beyond food to alcohol and cannabis cue-reactivity, and whether Pavlovian endophenotyping can serve as a practical stratification tool for adaptive digital interventions.

All studies in this programme adopt the Registered Report publication format, ensuring methodological rigour and publication regardless of outcome direction. Our validated gamified training platform, adopted by independent research groups internationally, supports the large-scale, app-based data collection required to characterise individual differences with adequate statistical power.

A structured translational pathway exists through BeweLab SA (bewe.com), a University of Fribourg spin-off, ensuring that validated endophenotyping procedures can be integrated into scalable digital health products under formal conflict-of-interest management between the university and the company.

Selected references from the lab

Tapparel, M. et al. (in press). Sign-tracking bias moderates Go/NoGo training-induced food devaluation. Registered Report.

Najberg, H. et al. (2023). Effects of gamified Go/NoGo training on sugar-sweetened beverage consumption. Registered Report.

Najberg, H. et al. (2021). A gamified inhibition training platform for food cue devaluation. Registered Report.

Hartmann, L. et al. (2016). [Electrical neuroimaging methodology.]

De Pretto, M. et al. (2019). [Topographic EEG and executive control.]

Sallard, E. et al. (2018). [Inhibitory control and neural plasticity.]