16 May 2018: Psychophysiological modelling is an emerging technique to analyse peripheral physiology in behavioural research (SCR, ECG, respiration, pupil, breathing, EMG, …). A new review now summarises the theoretical and statistical background, as well as all methods that have been developed by ourselves and other groups (Bach et al., Psychophysiology, 2018).

06 April 2018: PsPM 4.0.2 is released with further minor bug fixes.

09 March 2018: Giuseppe Castegnetti successfully defends his PhD thesis. Congratulations!

07 March 2018: We are running a PsPM workshop a the Meeting of the German Psychological Society/Section Biological Psychology (Psychologie & Gehirn). Information and registration can be found here.

23 February 2018: PsPM 4.0.1 is released with minor bug fixes.

01 February 2018: Dominik Bach receives Pfizer Research Award for the discovery that the antibiotic doxycycline, an inhibitor of eukaryot matrix metalloproteinases, attenuates fear memory in humans.

27 December 2017: Survival-relevant decisions include avoiding starvation, and this mandates planning to maintain homeostasis across extended time horizons. In a new study in Nature Communications, Christoph Korn shows how such planning in humans combines a normatively optimal strategy with a myopic heuristic, and how discrepancies between these two control strategies are being processed neurally (Korn & Bach, Nature Communications, 2018).

04 December 2017: Matthias Staib successfully defends his PhD thesis. Congratulations!

26 November 2017: PsPM 4.0 released – including startle eye blink EMG modelling and pre-processing, and many improvements for data import and pre-processing. Thanks to everybody involved, and in particular Tobias Moser! Get the software here.

03 November 2017: Learning to predict threat requires amygdala plasticity. Sensory cortices can also be crucially required, but their role in fear acquisition remains unclear. In a recent study in Neuroimage, PhD student Matthias Staib demonstrates that Heschl’s Gyrus, including A1, represents threat predictions during fear conditioning, equally for simple and complex auditory CS. Strikingly, the representations of CS+ and CS-, respectively, are similar for simple and complex sounds, and thus invariant from the actual sound properties, likely ruling out receptive field plasticity as underlying mechanism (Staib & Bach, Neuroimage, 2017).

03 August 2017: Model-based analysis of skin conductance responses (SCR) relies on assumptions on the mapping from sudomotor nerve activity (SNA) to SCR; all existing approaches model this as a linear time-invariant (LTI) system. Together with collaborating physiologists Barbara Namer and Mikael Elam, Samuel Gerster uses intraneural recording and stimulation techniques to investigates under what conditions these assumptions are met. Indeed, when stimulation frequency is below 0.6 Hz, more than 90% of signal variability is explained under an LTI model (Gerster S, Namer B, Elam M, Bach DR, Psychophysiology 2017).

05 May 2017: Hippocampal oscillations are frequently reported in rodent approach-avoidance conflict tests of anxiety, but their expression and function in humans is not understood. In a new paper by PhD student Saurabh Khemka, we use MEG to show that hippocampal gamma power relates to learned threat probabilities under human approach-avoidance conflict. Gamma power is theta-modulated, and the hippocampus synchronises with prefrontal cortex in theta band (Khemka S, Barnes G, Dolan RJ, Bach DR, Journal of Neuroscience 2017).

28 April 2017: Amygdala-thalamus and amygdala-cortex connections are crucial for many cognitive operations, but our knowledge on these often comes from non-human research. In a new paper by Aslan Abivardi & Dominik Bach, we provide detailed quantitative in vivo analyses of white-matter connectivity between amygdala and thalamus subnuclei as well as cortical parcellations (Abivardi A & Bach DR, Human Brain Mapping 2017). A probabilistic atlas of these connections is available online here.



04 April 2017: Extracellular matrix (ECM) proteins may play a fundamental role for memory (LTP) formation, in vitro research suggests. Yet, there is little evidence from living animals, non-human or human. Strikingly, the simple antibiotic doxycycline inhibits matrix metalloproteinase (MMP) 9, which is critical for LTP, and doxycycline has been used to treat ECM disease outside the brain for at least a decade. This made it possible to test in humans to assess the impact of this MMP-9 inhibitor on fear conditioning. In a study published today, we find that a single dose of 200 mg doxycycline, ingested before fear acquisition, reduces fear retention 1 week later by around 60%, as indexed by fear-potentiated startle. This makes the drug potentially interesting for prevention or treatment of posttraumatic stress disorder (PTSD) (Bach DR, Tzovara A, Vunder J, Molecular Psychiatry 2017).

See also: UZH press release (German version), UCL press release, media coverage on, Reuters

31 March 2017: Emotions have fascinated philosophers and scientists for millennia; yet there is even little consensus on what phenomena they encompass, let alone on their constituent psychological and neural mechanisms. In an opinion article for Nature Reviews Neuroscience, Dominik Bach and Peter Dayan build on decision theory to analyse emotions and reconcile diverging theories  (Bach DR & Dayan P, Nature Reviews Neuroscience 2017)

03 February 2017: During anticipation of a probabilistic reward, where and how are statistics of the reward distribution computed? In our latest paper, we use MEG to address the temporal evolution of this encoding, both for expected reward and its variability. Beyond regions previously known to encode such statistics (such as OFC), we find a surprisingly early and robust encoding in visual and polymodal parietal areas, unexplained by visual features of the predictive stimuli (Bach DR, Symmonds M, Barnes G, Dolan RJ, in press, Journal of Neuroscience).


02 February 2017: Approach/avoidance conflict has been proposed as experimental model of human anxiety-like behaviour, and is sensitive to hippocampus lesions as in animals. In a new paper, Christoph Korn and colleagues show that this model is also sensitive to the anxiolytic drug lorazepam, thus highlighting the validity of the set up. Furthermore, by investigating two rare patients with Urbach-Wiethe disease, they find that selective amygdala lesions and hippocampus have a similar impact on the game, acting like a strong anxiolytic (Korn CK, Vunder J, Miro J, Fuentemilla L, Hurlemann R, Bach DR, in press, Biological Psychiatry).

19 January 2017: For us, 2016 was the year of Psychophysiological Modelling – with 7 research papers on methods & models development. In 2017, we will demonstrate how these methods (and our ethobehavioural computer games) can be used to tackle neuroscience questions.

08 November 2016: We are hiring a post doc – more info here.

03 November 2016: The pupil provides a window on cognition. To quantify fear learning from pupil dilation, we have now developed a new model-based approach. It offers higher precision than most other psychophysiological methods, including SCR. The model is built and validated on a range of data sets. This furnishes another novel and useful tool for the study of human Pavlovian fear conditioning (Korn CK, Staib M, Tzovara A, Castegnetti G, Bach DR, in press, Psychophysiology).

25 October 2016: PsPM 3.1 is released. This is the first version that truly deserves the name “Psychophysiological Modelling” and comprises model-based methods for analysis of heart rate data, respiration data, and pupil size. Furthermore, it includes convenience functions for these data and for startle eye-blink EMG analysis. Many minor improvements, including more generous handling of missing values, support for Philips Scanphyslog files, and for bioread-converted AcqKnowledge files. Download

20 October 2016: Bachlab is on Twitter – @bachlab_uzh.

28 September 2016: MRI scanners are commonly equipped with respiration chest belts for artefact correction. It is now possible to use these data to probe Pavlovian fear memory. In our new study, we demonstrate that respiratory amplitude responses (RAR) differentiate between CS+ and CS- in Pavlovian cued fear conditioning. We then develop and optimise a psychophysiological model (PsPM) to detect CS+/CS- differences in RAR. Comparing this method with SCR and HPR, we find that RAR performs similar to SCR, but is less sensitive than HPR, in inferring fear learning. Overall, this work provides a novel and robust tool to study fear memory in humans (Castegnetti G, Tzovara A, Staib M, Gerster S, & Bach DR, in press, Psychophysiology).

14 September 2016: Many researchers use fear-potentiated startle to make statements about fear learning, but a multiplicity of analysis approaches means that results may not be comparable across laboratories. In Saurabh’s and Athina’s new paper, they compare various analysis methods for human startle eyeblink response (SEBR) and develop a novel approach in the framework of Psychophysiological Modelling (PsPM). They further show that SEBR allows much more sensitive inference on fear learning than SCR or HPR (Khemka S*, Tzovara A*, Gerster S, Quednow BB, Bach DR, in press, Psychophysiology, * equal contribution).

31 August 2016: Lab BBQ and farewell to Saurabh, Johanna, & Jennifer. All the best for you!

Lab BBQ 2016

06 June 2016: Behavioural inhibition is a common observation under anxiety. We show that it is likely under instrumental control rather than Pavlovian. An explicit model that would make behavioural inhibition adaptive is also expressed in non-anxiety tasks, highlighting a possibility that anxiety-like behavioural inhibition is model-based (Bach DR 2016. Journal of Experimental Psychology: Human Perception and Performance).

02 June 2016: Respiration belts are common equipment in MRI scanners, and used to correct fMRI image time series for motion artefacts. In a recent paper, we present a method to infer cognitive states from respiration measures, based on the simple systems employed in MRI environments. This method builds on our PsPM framework and uses a general linear convolution model on interpolated respiration period, amplitude, and respiratory flow rate. We are now working towards using this method to infer fear memory from respiration (Bach DR, Gerster S, Tzovara A, Castegnetti G, 2016, Journal of Neuroscience Methods).

18 May 2016: The Symposium on Advances in Computational Neuroimaging, organised by University of Zurich and University College London, is a great success. Below are some pictures. Detailed program can be found here, UZH press coverage here. Lab member Christoph Korn gave an engaging talk on action control during foraging under threat of starvation.

An attentive audience at UCL (Photos Adrian Ritter, UZH).

An impressive speaker line up (Photo Adrian Ritter, UZH).

Klaas Enno Stephan (Keynote speaker) and Dominik R Bach (Organiser) (Potos Adrian Ritter, UZH).

19 March 2016: We are recruiting a graduate student on a funded PhD position, working on computational neuroimaging of human fear conditioning. Find more information here.

10 February 2016: Heart period response (HPR) to threatening stimuli is characterised by a phasic increase, termed fear bradycardia. Its reliance on non-invasive measures (e.g. ECG) makes it suitable for assessing associative fear memory in humans. We thus developed a novel PsPM that, when applied to the analysis of HPR from fear conditioning experiments, robustly distinguishes between responses to fear conditioned and neutral stimuli, and does so better than any other classical scoring method in the literature (Castegnetti G, Tzovara A, Staib M, Paulus PC, Hofer N, & Bach DR, 2016, Psychophysiology).

20 Januar 2016: Pupil size offers a window on cognition. Here, we attempt to create a Psychophysiological model (PsPM), to infer cognitive processes from pupil responses. We find that pupil responses to illuminance changes can be modelled by a combination of two LTI systems. Since illuminance-related and cognitive inputs are presumed to share a common final pathway, we can use our illuminance-based model to estimate the temporal evolution of the neural inputs for cognitive tasks (e.g., an auditory oddball task, an emotional words task, and a visual detection task) (Korn CW & Bach DR, 2016, Journal of Vision).

05 Januar 2016: Cardiac rhythm is influenced by autonomic input, and this input may be inferred from measured heart period. We harness the PsPM framework to build a quantitative model for event-related heart period responses (HPR). This model allows estimating the amplitude of separable HPR components from measured ECG, and these may be interpreted as autonomic input components. Ongoing work investigates how this model can be finessed to infer fear memory from HPR (Paulus PC, Castegnetti G, & Bach DR, 2016, Psychophysiology).

01 December 2015: PsPM 3.0.2 is released with minor bug fixes.  Download

17 November 2015: 2nd Burghölzli Psychiatry Meeting

04 November 2015: Behavioural inhibition is a key anxiety behaviour in humans, non-human primates, and rodents. Its functional role is however unclear – it might arise from decision difficulty, or fulfil one of various adaptive goals. In a mathematical model, we show that behavioural inhibition can optimally balance survival and foraging success under environmental threat-reward correlations. Empirically, humans show behavioural inhibition in an approach/avoidance task and is not due to decision difficulty or risk assessment behaviour (Bach DR 2015, PLoS Computational Biology).

07 August 2015: Fear memory can be quantified using skin conductance responses (SCR), by inverting a Psychophysiological Model (PsPM) formulated in the framework of Dynamic Causal Modelling (DCM). In a detailed methodological investigation, this approach is validated and optimised, in order to maximise predictive validity (Staib M, Castegnetti G, & Bach DR 2015, Journal of Neuroscience Methods, 255, 131-138).

04 June 2015: PsPM 3.0.1 is released. It provides platform-independent import of WDQ-files, EDF import, convenience functions for respiration channels, z-scoring first-level statistics for contrasts, and many minor bug fixes. Stay tuned for the next version which will support new data modalities. Download

27 April 2015: Maintaining homeostasis is a fundamental requirement for living organisms. We show that a novel model based on homeostatic principles is better suited to explain human choice data than standard economic models – both in a virtual foraging task and in casino-like risky gambles  (Korn CW & Bach DR 2015, PLoS Computational Biology).

06 February 2015: PsPM is a powerful matlab toolbox for model-based analysis of psychophysiological signals, for example SCR, ECG, EMG, respiration, or pupil size. Today, we release a new version, PsPM 3.0. It offers an entirely novel GUI, building on the Matlab Batch Editor well known to SPM users. We worked > 1 year on streamlining the code, testing basic functions, fixing bugs, and enhancing usability. A tutorial now offers a smooth introduction into usage of the software. Get the software here.

03 February 2015: Matching pursuit is a machine-learning algorithm that allows for fast model inversion. This algorithm can be used to infer tonic arousal from skin conductance recordings – by estimating the number of sudomotor bursts. This estimation is about 100 times faster than the currently available DCM inversion for this purpose, and not less precise in terms of predictive validity (Bach DR & Staib M 2015, Psychophysiology)

27 January 2015: Startle reflex is modulated by state of the environment, e.g. in fear-potentiated startle. A pattern of modulation previously unexplained is observed during anticipation of reward, or instructed attention. A Bayesian inference and cost-minimisation model predicts this pattern, including fear-potentiation, from first principles (Bach DR, Journal of Theoretical Biology 2015, 370, 53-60).

22 October 2014: 1st Burghölzli Psychiatry Meeting

05 October 2014: Auditory reward-predicting stimuli enhance visual detection, and this is also apparent in neural signal in visual cortex. This provides evidence for a value-driven cross-modal interaction that affects both perception and stimulus encoding (Pooresmaeili, FitzGerald, Toelch, Bach, Ostendorf, & Dolan, PNAS, 111, 15244-15429)

19 August 2014: The amygdala is thought crucial for rapid threat detection, but findings from human lesion studies are ambiguous. We show that two patients with selective bilateral amygdala lesions are impaired in prioritising resources towards angry rather than happy facial expression (Bach, Hurlemann, & Dolan, Cortex, 63, 206-213)

11 August 2014: SCRalyze provides a model-based approach to analysing skin conductance responses (SCR). In a head-to-head comparison with an alternative approach engendered in the software Ledalab, SCRalyze performs significantly better in almost all comparisons and comparable in the remaining ones (Bach DR, Biological Psychology, 103, 63-88).

22 July 2014: Emotional facial expression enhances identity recognition of the same face and attentuates identity recognition of other faces, providing evidence that attention to emotional information in faces is based on objects, not features (Bach, Schmidt-Daffy, & Dolan, Emotion, 14, 1007-1013).

4 March 2014: Animal models of human anxiety often invoke a conflict between approach and avoidance. In a combined functional magnetic resonance imaging and hippocampus lesion study, we show that the human anterior hippocampus is engaged in monitoring threat level during approach-avoidance conflict. This replicates and extends a rodent literature on models of anxiety and allows cross-species investigation of these models (Bach, Guitart-Masip, et al., Current Biology, 24, 541-547).