Area 3: Computational Psychiatry and Aging

Project 3.1: Worry, Anxiety and Perception of Danger Throughout the COVID-19 Pandemic: A Longitudinal Study

Project 3.2: Effect of L-DOPA on Neural Signals of Walking Direction in Older and Younger Adults

A third focus of our group is developing and testing computational and neural theories of learning and decision making in psychiatric disorders and healthy aging. Our efforts relate to recent work showing the value of mathematical modeling in addressing both issues and reflects our group’s involvement in the Max Planck UCL Centre for Computational Psychiatry and Aging Research. In the domain of psychiatry, the group’s research has focused on the role of anxiety in aversive learning (see Zika et al., 2023 and Project 7 below). Our findings reveal that anxious individuals are better at building higher-order representations (a simple form of cognitive maps) and our goal for the future is to connect our work on anxiety with cognitive maps and replay, the main topics of Areas 1 and 2. In our work on aging we have recently investigated how older age affects the neural representations that underlie spatial navigation (Koch et al., 2020), and which role dopamine plays in representational changes (see Project 3.2 below; Koch et al., 2022). More recently, we used reinforcement learning models to investigate factors impacting learning from surprising events in young and older adults (Koch et al., 2023).

Key References

Gagne, C., Zika, O., Dayan, P., & Bishop, S. J. (2020). Impaired adaptation of learning to contingency volatility in internalizing psychopathology. eLife, 9, Article e61387.
Koch, C., Baeuchl, C., Glöckner, F., Riedel, P., Petzold, J., Smolka, M. N., Li, S.-C., & Schuck, N. W. (2022). L-DOPA enhances neural direction signals in younger and older adults. NeuroImage, 264, Article 119670.
Koch, C., Zika, O., & Schuck, N. W. (2022). Influence of surprise on reinforcement learning in younger and older adults. PsyArXiv, December 2, 2022.
Zika, O., Wiech, K., Reinecke, A., Browning, M., & Schuck, N. W. (2023). Trait anxiety is associated with hidden state inference during aversive reversal learning. Nature Communications, 14, Article 4203.

Project 3.1: Worry, Anxiety and Perception of Danger Throughout the COVID-19 Pandemic: A Longitudinal Study 

Research Scientists
Ondrej Zika
Claire Gillan (Trinity College Dublin, Ireland)
Nicolas W. Schuck

In this project we investigated how worry and perception of threat changed over the course of the COVID-19 pandemic, especially in relation to individual differences in trait anxiety and depression. Our research links to our own and others’ previous work associating anxiety with increased sensitivity to underlying changes in contextual threat (Zika et al., 2023). While past work in the area has mostly relied on controlled laboratory experiments with limited ecological validity, our study was situated within a natural real-life occurrence of a fluctuating threat, the COVID-19 pandemic, and hence offers unique ecological validity.

Participants from the UK and Germany (n = 400) completed a series of online questionnaires over 20 measurement time points between April and December 2020. Participants answered questions relating to health and economic worry, aversive probability estimates (e.g., “What’s the probability of infection/death due to COVID-19?”) and perception of danger (i.e., “Are we currently in a period of danger?”) on each of the 20 testing time points. To index objective environmental threat, we additionally collected publically available statistics about new COVID-19 cases and deaths in each postcode and matched it to participants’ locations. Finally, individual differences in anxiety and depression as assessed by STAI-T, STICSA-T, and BDI-III questionnaires were projected onto three distinct factors: cognitive anxiety and depression (TF1), somatic anxiety (TF2), and unhappiness (TF3), using a factor analysis.

First, we investigated how perceived anger and worry relate to the recorded objective levels of threat (e.g., deaths). Figure 6a shows the timecourses of the objective and subjective measures. While all subjective measures were elevated in the first wave of the pandemic (i.e., spring 2020), only the perceived danger increased during the second wave (late summer 2020). On the other hand, worries and infection probability estimates remained unchanged despite the second surge in COVID-19 cases, highlighting a dissociation between worry, perceived danger, and objective threat. Next, we asked how the trait factors related to the evolution of worry, perceived danger, and infection probability estimate. Interestingly, somatic anxiety (TF2) was found to positively predict all three, but cognitive anxiety (TF1) was not. We followed up on this finding by running a temporal clustering analysis. Here, the individuals who more consistently perceived high danger throughout the entire pandemic were significantly higher in somatic trait anxiety compared to those who more adaptively changed their perception of threat as cases and deaths fell and later increased (Figure 6b). Breaking down worry into health and economic components revealed that somatic anxiety was associated with health but not economic worry. Unhappiness, on the other hand, was associated with both (Figure 7). Finally, we looked at the relationship between trait factors and objective severity, indexed by number of deaths. While both somatic and cognitive anxiety were not related to the number of COVID-19-related deaths, unhappiness was positively correlated (r = 0.31). In summary, we found that high somatic but not cognitive anxiety was associated with increased worry, aversive probability estimates, and perception of danger, even in times of relative safety. Specifically, somatic anxiety was associated with health-related but not economic worry. Despite higher levels of worry, objective threat was not elevated in high somatic anxiety. Taken together, these results suggest that during real world threat individuals high in somatic anxiety experience higher levels of worry irrespective of the objective level of threat in the environment. In contrast, unhappiness and resulting worry in some individuals can partially be explained by higher objective threat of the environment.

Project 3.2: Effect of L-DOPA on Neural Signals of Walking Direction in Older and Younger Adults

Research Scientists
Christoph Koch
Christian Baeuchl (Technische Universität Dresden, Germany)
Franka Glöckner (Technische Universität Dresden, Germany)
Philipp Riedel (Technische Universität Dresden, Germany)
Johannes Petzold (Technische Universität Dresden, Germany)
Michael Smolka (Technische Universität Dresden, Germany)
Shu-Chen Li (Technische Universität Dresden, Germany)
Nicolas W. Schuck

Aging is characterized by vast declines in spatial navigation ability and dopaminergic neuromodulation. How are these factors related? Project 8 extended our previous work on walking direction representations in older versus younger adults (Koch et al., 2020) by investigating the effect of L-DOPA administration on these neural signals. Building on computational models that have proposed that deficits in dopamine functioning negatively affect cognition through losses in the signal-to-noise ratio of neural responses, we expected dopamine to affect the specificity of neural representations of spatial environments. Our previous work has shown that without L-DOPA interventions, older brains tend to have more dedifferentiated neural representations of virtual walking direction. Here, we therefore asked if DA causally influences the specificity of older and younger adults’ neural representations in a spatial task.

We tested the influence of DA on the specificity of neural representations of walking direction. In a double-blind drug intervention design, 37 older and 43 younger adults completed a spatial memory task while undergoing fMRI. In one drug (L-DOPA) and one placebo session participants navigated a 3D virtual environment while learning and remembering locations. The paths traveled during the task were classified by walking directions and the associated fMRI activation patterns were extracted. Neural specificity was then measured by how accurately a cross-validated multivariate classifier could identify walking directions from neural patterns. Comparing this measure across sessions quantified the causal influence of DA on neural specificity.

Walking direction could be predicted above-chance in brain areas involved in spatial navigation, including the early visual cortex, retrosplenial cortex (RSC), and hippocampus. Quantifying the effect of DA, a statistical model revealed increased specificity/decoding of walking direction representations in the drug versus placebo session. Exploratory follow-up analyses revealed that the increase was strongest in the hippocampus, where decoding was only possible under higher levels of DA. Furthermore, while the drug increased neural specificity in both age groups in the hippocampus, in the RSC this was unique to younger adults. These findings provide first evidence for a causal link between DA and the specificity of neural responses during spatial navigation. While the age-dependent effects in the RSC raise questions about a more complex role of DA in the aging navigational system, our findings shed light on DA’s contribution to the pronounced decline of spatial cognition in older age.

Go to Editor View