New Results

Catecholamines reduce choice history biases

Jan Willem de Gee, Niels A. Kloosterman, Anke Braun, Tobias H. Donner
doi: https://doi.org/10.1101/2024.10.28.620689

Abstract

Theoretical accounts postulate that the catecholaminergic neuromodulator noradrenaline shapes cognitive behavior by reducing the impact of prior expectations on learning, inference, and decision-making. A ubiquitous effect of dynamic priors on perceptual decisions under uncertainty is choice history bias: the tendency to systematically repeat, or alternate, previous choices, even when stimulus categories are presented in a random sequence. Here, we directly test for a causal impact of catecholamines on these priors. We pharmacologically elevated catecholamine levels through the application of the noradrenaline reuptake inhibitor atomoxetine. We quantified the resulting changes in observers’ history biases in a visual perceptual decision task. Choice history biases in this task were highly idiosyncratic, tending toward choice repetition or alternation in different individuals. Atomoxetine decreased these biases (toward either repetition or alternation) compared to placebo. Behavioral modeling indicates that this bias reduction was due to a reduced bias in the accumulation of sensory evidence, rather than of the starting point of the accumulation process. Atomoxetine had no significant effect on other behavioral measures tested, including response time and choice accuracy. We conclude that catecholamines reduce the impact of a specific form of prior on perceptual decisions.

Introduction

Biases are prevalent in human decisions ranging from those based on abstract reasoning to judgments of the sensory environment1–3. One prominent source of such biases in perceptual decisions across mammalian species is the history of previous experiences4–13. When, as common in natural environments, the environmental state evaluated in the judgment is relatively stable, such history biases reflect adaptive expectations that are dynamically updated over time and improve performance5,14,15. For example, when choosing to forage for deer in the same area as during successful previous expeditions. While recent work in rodents and primates has identified correlates of such history biases in neural population activity of parietal and frontal cortical areas9,12,16–20, little is currently known about the neurotransmitter systems involved in this important form of bias.

Influential theoretical work postulates that the catecholaminergic neuromodulator noradrenaline shapes behavior by reducing the impact of prior expectations on learning, inference, and decision-making21,22. The catecholaminergic (and in particular, the noradrenaline) systems of the brainstem project to large parts of the cerebral cortex, where the neuromodulators they release change the functional properties of their target networks23–28. Consequently, they are in an ideal position to shape cortex-wide network activity underlying decision-making in a coordinated fashion. Indeed, pupil responses (a peripheral proxy of central arousal state29 point to a role of neuromodulation in regulating the impact of these history-dependent priors8. However, pupil responses reflect the activity of multiple neuromodulatory brainstem systems29–31, and the above result builds on correlative approaches. Direct evidence for the role of catecholaminergic neuromodulation in choice history biases is currently sparse in animal models13,32 and absent in humans.

Here, we aimed to provide causal evidence for the role of catecholamines in regulating (suppressing) history biases in human choice behavior. Specifically, we hypothesized that catecholamines down-regulate the impact of previous experiences on decision-making, which should translate into a reduction of choice history bias. We tested this hypothesis by pharmacologically increasing central catecholamine (in particular noradrenaline) levels through the selective noradrenaline re-uptake inhibitor atomoxetine33,34 and quantified the resulting behavioral changes in a variant of a visual perceptual decision task adopted from recent animal physiology35. Atomoxetine inhibits the presynaptic norepinephrine transporter, preventing the reuptake of norepinephrine throughout the brain along with inhibiting the reuptake of dopamine in specific brain regions such as the prefrontal cortex33,34. We found that atomoxetine reduced choice history biases by specifically interacting with the way people accumulated noisy perceptual evidence.

Results

Atomoxetine was administered orally (40 mg per session) in a within-subject, double-blind, placebo-controlled, and randomized design (Figure 1A). The behavioral task entailed the presentation of visual targets (Gabor patch of fixed orientation and varying contrast), one in the left and one in the right visual hemifield; participants were asked to report the location the larger contrast target by button press (Figure 1B; Methods). To increase sensory uncertainty and promote integration of evidence across time36, we added dynamic visual noise to both Gabor patches (Figure 1B; Methods). There were two different signal strengths (small and large difference in contrast); auditory feedback was presented 50 ms right after participant’s choices. We collected many trials per participant (range, 3520-4320) across four experimental sessions, in which we systematically varied and counterbalanced drug condition and stimulus-response mapping (Figure 1A and Figure S1; Methods).

Figure 1.
Figure 1. Experimental design and effects on peripheral arousal markers.

(A) Design was within-subject, double-blinded, randomized, and placebo-controlled (Methods). Additionally, the stimulus-response mapping was counterbalanced within participants (Figure S1; Methods). (B) Schematic sequence of events during the spatial contrast discrimination task (Methods). Participants reported the location of the highest contrast Gabor patch (difference in contrast is high for illustration only). (C) Heart rate (Methods), separately for placebo and atomoxetine sessions. Every connecting line is a participant; large data points in the middle are the group averages (error bars, s.e.m. across 19 participants); stats, paired-sample t-test. (D) As C, but for pupil size (Methods).

Atomoxetine administration at this relatively low dose had a measurable effect on peripheral arousal markers, but not on the overall efficiency of the decision process for the behavioral task at hand. Atomoxetine increased both heart rate (Figure 1C) and pupil size (Figure 1D). While signal strength lawfully decreased reaction time (RT; F1,18 = 16.2, p = 0.001; two-way repeated measures ANOVA) and increased accuracy (F1,18 = 297.3, p < 0.001) (Figure 2A,B), atomoxetine changed neither of those behavioral parameters (RT: F1,18 = 2.3, p = 0.146; accuracy: F1,18 = 0.8, p = 0.383; no significant interaction of signal strength and drug; Figure 2A,B). Likewise, atomoxetine did not significantly change signal detection theoretic sensitivity (d’; Figure S2A).

Figure 2.
Figure 2. Atomoxetine reduces choice history bias.

(A) Reaction time, separately for placebo and atomoxetine sessions, and separately or weak and strong signals. Every connecting line is a participant; large data points in the middle are the group averages (error bars, s.e.m. across 19 participants); stats, main effect of drug in two-way repeated measures ANOVA. See Figure S2B for reaction time distributions. (B) As A, but for accuracy. (C) Probability of repeating the previous stimulus category (left) or choice (right) during placebo sessions. (D) As A, but for repetition probability (Methods). (E) Individual shift in repetition probability caused by atomoxetine, plotted against individual’s repetition probability during the placebo sessions. Data points are individual participants; error bars, 68% confidence intervals across 5K bootstraps; stats, Pearson’s correlation coefficient (corrected for reversion to the mean; Figure S2C; Methods). (F) As A, but for absolute repetition probability (Methods).

By contrast, atomoxetine had a robust impact on the impact of history-based (i.e., dynamically varying throughout the course of the experiment) prior expectations on their decisions. We operationalized these time-varying prior expectations about the next target as the probability of repeating the choice from the previous trial (Figure 2C). The stimulus repetition probability was, by design, close to 0.5 (Figure 2C) (group average: 0.496; Methods), thus not explaining the sequential effects in participants’ behavior. Atomoxetine reduced the mean choice repetition probability (Figure 2D; F1,18 = 8.2, p = 0.010; two-way repeated measures ANOVA). Signal strength had no such effect (F1,18 = 0.1, p = 0.706) and signal strength and drug did not interact (F1,27 = 0.5, p = 0.507).

In the placebo sessions, most participants (15 out of 19) had a repetition probability larger than 0.5, that is, they tended to repeat their choices (Figure 2C). We, therefore, wondered if (i) atomoxetine reduced the group-level choice history biases by stereotypically pushing behavior more toward alternation (i.e., reduction of repetition probability, in line with previous pupillometry work8) or (ii) through a reduction of both repetition and alternation biases (shift of repetition probability toward 0.5). Our results are in line with the second scenario. There was a robust correlation between individual choice history biases during placebo sessions and the shift in history biases caused by atomoxetine: participants with the strongest repetition or alternation biases, exhibited the strongest shift towards a repetition probability of 0.5 (Figure 2E; corrected for reversion to the mean; Methods). Consequently, atomoxetine yielded an even more consistent and robust reduction of the absolute choice history bias, defined as the absolute value of P(repeat)−0.5 (Figure 2F; F1,18 = 10.6, p < 0.001; two-way repeated measures ANOVA). Atomoxetine did not change overall choice bias, quantified as signal detection theoretic criterion (Figure S2D-F). These findings show that atomoxetine specifically reduced the impact of history-dependent prior expectation in all participants, regardless of whether they idiosyncratically tended to repeat (i.e., assuming a relatively stable environment) or alternate (assuming a volatile environment).

How did participants’ expectations shape their decision processes, and how did atomoxetine act on this mechanism? Current models of perceptual decision-making posit the temporal accumulation of sensory evidence, resulting in an internal decision variable that grows with time36–39. When this decision variable reaches one of two decision bounds, a choice is made, and the corresponding motor response is initiated. In this framework (Figure 3A), a bias can arise in two ways: (i) by shifting the starting point of accumulation toward one of the bounds or (ii) by changing the rate of evidence accumulation toward one choice alternative.

Figure 3.
Figure 3. Atomoxetine reduces choice history bias in evidence accumulation.

(A) Schematic of drift diffusion model accounting for choices, and their associated RTs. In the equation, v is the drift rate. Purple and cyan curves show expected RT distributions under shifts in either the ‘starting point’ (z; purple) or ‘drift bias’ (vbias; cyan). (B) Left: history-dependent shift in starting point bias (Methods), separately for placebo and atomoxetine sessions. Right, as left, but for drift bias. Every connecting line is a participant; large data points in the middle are the group averages (error bars, s.e.m. across 19 participants); stats, paired-sample t-test. (C) Left: individual shift in repetition probability caused by atomoxetine, plotted against individual’s history-dependent shift in starting point bias. ‘History shift’ is the difference between a parameter estimate for previous ‘left’ and previous ‘right’ choices, irrespective of the category of the current stimulus. Right: as left, but for history-dependent shift in drift point bias. Data points are individual participants; stats, Pearson’s correlation coefficient (difference in correlation assessed with permutation test; Methods).

Previous work has shown that individual biases in choice repetition probabilities are generally explained by a history-dependent bias in the drift rather than starting point, across a range of different perceptual choice tasks4. We replicated this finding for the placebo condition of the present dataset (Figure S3A). To this end, we fitted an accumulation-to-bound model of decision-making to the behavioral data. The model had five free parameters: 1) boundary separation (controlling response caution); 2) the mean drift rate (overall efficiency of evidence accumulation); 3) non-decision time (the speed of pre-decisional evidence encoding and post-decisional translation of choice into motor response); 4) the starting point of the decision; 5) an evidence-independent bias in the drift (henceforth called “drift bias”). We fitted all five parameters separately per drug condition; additionally, we fitted drift rate separately for both signal strength conditions, and both starting point and drift bias separately for the previous choice category (‘left’ vs ‘right’). The fitted model (Figure S3C) accounted well for the overall behavior in each task (Figure S3D,E). Indeed, atomoxetine also specifically reduced the history-dependent shift in drift bias, but not the history shift in starting point (Figure 3B). Furthermore, this drift bias (not starting point) effect was robustly and specifically correlated to the individual change of choice repetition probability (Figure 3C).

Discussion

Using selective pharmacological interventions in humans, we here imply catecholamines in the modulation of human choice history biases: boosting catecholamines levels through atomoxetine specifically reduced the magnitude of choice history biases, regardless of whether those promoted choice repetition or alternation. We interpret this finding in the context of computational theory postulating a down-weighting of prior expectations (which, we assume, drive the choice history biases analyzed here) in combination with new sensory evidence in the evaluation of current environmental state21,22. Catecholamines have previously been implicated in modulating learning rates40–43 and facilitating set shifting44 in dynamic environments; here, we provide causal evidence for their role in controlling the weight of historical information in perceptual decisions.

Our finding is roughly consistent with previous correlative evidence from pupillometry8,45–49. Non luminance-mediated changes of pupil diameter are an indirect marker of neuromodulatory (including noradrenergic and cholinergic) activity30,31,50–54. In dynamic environments, pupil responses are associated with violations of learned, top-down expectations and reduce their weight45–49. In stable environments, pupil responses predict a reduction of choice repetition probability8. There are, however, also notable differences between our current results and this previous pupillometry work. First, the pupil-predicted reduction of repetition probability was due to a stereotypical increase in the tendency to alternate8. In contrast, we here found that catecholamines reduce both repetition and alternation biases (shift of repetition probability toward 0.5), which is more readily in line with a reduction in prior weight. Second, our pharmacological intervention did not reduce participant’s overall choice bias (bias towards one choice alternative, irrespective of the immediate history), even though this is another commonly reported behavioral correlate of pupil responses31,55–58. One possible explanation for this difference is that atomoxetine likely increased both tonic and phasic catecholamines levels59, while task-evoked pupil responses reflect phasic neuromodulatory activity31. Another possible explanation is that while atomoxetine selectively elevates central catecholamine levels (noradrenaline and dopamine), pupil responses track not only noradrenaline30,31,50–53, but also acetylcholine30,54, serotonin60 and orexin61 levels.

Which neuromodulator specifically mediated the effects reported here? Atomoxetine increases the extracellular levels of noradrenaline, but also of (prefrontal) dopamine (Methods). Indeed, one recent report suggest that striatal dopamine may play a role in adapting choice history bias to temporal regularities in a stimulus sequence13. Thus, our results are consistent with a role of one or both catecholamines. However, the noradrenergic locus coeruleus and dopaminergic midbrain exhibit a close interplay62,63 and are co-activated with pupil responses during decisions31. Thus, it could well be that their (complex) interaction regulates choice history bias in perceptual decisions.

In sum, pharmacologically elevating central catecholamine (specifically noradrenaline) levels in human participants specifically reduces their propensity to accumulate noisy evidence for perceptual decisions in a manner that is biased by previously reported perceptual experience. This supports computational theories according to which noradrenaline reduces the impact of dynamic priors on inference and decision-making.

Materials and Methods

Participants

Nineteen participants (14 female, age range 20-26) participated in the experiment, entailing concurrent pupillometry and MEG recordings. All had normal or corrected to normal vision and no history or indications of psychological or neurological disorders. The experiment was approved by the ethics committee of the University Medical Center Hamburg-Eppendorf. Participants participated in four experimental sessions, in which we crossed drug condition (atomoxetine vs. placebo) and stimulus-response mapping (press ipsi vs. contra) (about 2 hours per session). Participants were paid 10€ per hour. One participant completed three recording sessions.

Pharmacological intervention

We used the selective noradrenaline reuptake inhibitor atomoxetine (dose, 40 mg) to boost the levels of catecholamines, specifically noradrenaline and (in the prefrontal cortex) dopamine33,34. Atomoxetine is a relatively selective inhibitor of the noradrenaline transporter, which is responsible for the natural reuptake of noradrenaline that has been released into the extracellular space. Consequently, atomoxetine acts to increase the extracellular levels of noradrenaline, an effect that has been confirmed experimentally in rat prefrontal cortex33. The same study showed that atomoxetine also increases the prefrontal levels of dopamine, which has a molecular structure very similar to the one of noradrenaline and is, in fact, a direct precursor of noradrenaline. Atomoxetine has smaller affinity to the serotonin transporter, and there are discrepant reports about the quantitative relevance of these effects: while one study found no increases in serotonin levels under atomoxetine33, a recent study reports a significant atomoxetine-related occupancy of the serotonin transporter in nonhuman primates64 at dosages that would correspond to human dosages of 1.0– 1.8 mg/kg. Note that these dosages are substantially higher than the administered dosage in this study (40 mg, independent of body weight). It is therefore unclear to what extent our atomoxetine condition affected cortical serotonin levels. A mannitol-aerosil mixture was administered as placebo. All substances were encapsulated identically to render them visually indistinguishable. Peak plasma concentrations are reached about 1–2 hours after administration for atomoxetine, and the half-life is about 5 hours65. Thus, participants received either a placebo or atomoxetine 1.5 hours before the start of the experimental session.

Behavioral task

Each trial consisted of three consecutive intervals (Figure 1A): (i) the baseline interval (uniformly distributed between 0.6 and 1.4 s) containing a dynamic noise pattern on both the left and right side of the green fixation dot; (ii) the decision interval (terminated by the participant’s response; max 3 s) containing the same noise patterns with Gabor patch embedded in each; (iii) the feedback interval (uniformly distributed between 0.25 and 1.4 s).

Two dynamic noise patterns were presented throughout the experiment. The luminance across all pixels was kept constant. This pedestal noise pattern had 20% contrast. In the decision interval two Gabor patches (sinusoidal gratings; 2 cycles per degree; phase randomly drifted left or right) were added to each noise pattern. The Gabor patches symmetrically differed from 80% contrast. On half of trials, the Gabor patch on right side of the screen was of higher contrast (signal, right); on the other half the Gabor patch on the left side was of higher contrast (signal, left). Signal location was randomly selected on each trial, under the constraint that it would occur on 50% of the trials within each block of 160 trials. Stimuli were presented within two circular patches to the left and right of the central fixation cross (diameter, 15 degrees of visual angle; eccentricity, 12 degrees).

Participants were instructed to report the location of the signal, the highest contrast Gabor patch by pressing one of two response buttons with their left or right index finger as soon as they felt sufficiently confident (“free response paradigm”). We did not give specific instructions to tradeoff accuracy for speed (or vice versa). The mapping between perceptual choice and button press (e.g., “right” –> press right key; “left” –> press left key) was counterbalanced across the four recording sessions. Participants received auditory feedback (high tone, correct; low tone, error) 50 ms after they indicated their choice.

Signal strength, the difference in contrast between the left and right signals, was individually titrated before the main experiment to two difficulty levels that yielded about 70% and 85% correct choices, using an adaptive staircase procedure (Quest)66. In the placebo sessions, we measured a mean accuracy of 74.07 % correct (± 0.97 % s.e.m.) and 80.83 % correct (± 1.12 % s.e.m), for weak and strong signal strengths respectively.

Stimuli were back-projected on a transparent screen using a Sanyo PCL-XP51 projector with a resolution of 800×600 at 24 Hz. The screen was positioned 65 cm away from their eyes. The luminance profile was linearized by measuring and correcting for the systems gamma curve. A doubling of contrast values therefore also produced a doubling of luminance differences. During the first training session stimuli were presented on a VIEWPixx monitor with the same resolution and refresh rate (also linearized). To minimize any effect of light on pupil diameter, the overall luminance of the screen was held constant throughout the experiment.

Participants performed between 22 and 27 blocks (distributed over four sessions) yielding a total of 3520–4320 trials per participant. One participant performed a total of 17 blocks (distributed over three scanning sessions), yielding a total of 2720 trials.

Data acquisition

Recordings took place in a dimly lit magnetically shielded room. Magnetoencephalography (MEG) was recorded using a whole-head CTF 275 MEG system (CTF Systems Inc., Canada) at a sampling rate of 1200 Hz. In addition, eye movements and pupil diameter were recorded with an MEG-compatible EyeLink 1000 Long Range Mount system (SR Research, Osgoode, ON, Canada). Finally, electrocardiogram (ECG) as well as vertical and horizontal electrooculogram (EOG) were acquired using Ag/AgCl electrodes at a sampling rate of 1200 Hz. MEG data will be the focus of a different report.

Analysis of electrocardiogram data

ECG data were used to analyze average heart rate. We used an adaptive threshold to detect the R-peak of each QRS-complex in the ECG. Heart rate was then computed by dividing the total number of R-components by time.

Analysis of pupil data

All analyses were performed using custom-made Python scripts, unless stated otherwise.

Preprocessing

Periods of blinks and saccades were detected using the manufacturer’s standard algorithms with default settings. The remaining data analyses were performed using custom-made Python scripts. We applied to each pupil timeseries (i) linear interpolation of missing data due to blinks or other reasons (interpolation time window, from 150 ms before until 150 ms after missing data), (ii) low-pass filtering (third-order Butterworth, cut-off: 6 Hz), (iii) removal of pupil responses to blinks and to saccades, by first estimating these responses by means of deconvolution and then removing them from the pupil time series by means of multiple linear regression67, and (iv) conversion to units of modulation (percent signal change) around the mean of the pupil time series from each measurement session.

Quantification of pre-trial pupil size

We quantified pre-trial pupil size as the mean pupil size during the 0.25 s before trial onset.

Analysis and modeling of choice behavior

All analyses were performed using custom-made Python scripts, unless stated otherwise. The first trial of each block was excluded from the analyses. We always computed behavioral metrics separately for each participant and each of the four recording sessions, and then averaged over the two SR-mappings within each drug condition.

Overt behavior

Reaction time (RT) was defined as the time from stimulus offset until the button press. Repetition probability was defined as the fraction of choices that was the same as the choice on the previous trial.

Signal-detection theoretic modeling

The signal detection68 metrics sensitivity (d’) and criterion (c) were computed separately for each of the pupil bins. We estimated d’ as the difference between z-scores of hit-rates and false-alarm rates. We estimated criterion by averaging the z-scores of hit-rates and false-alarm rates and multiplying the result by -1.

Drift diffusion modeling

We fitted the reaction time data with the drift diffusion model37,39, based on continuous maximum likelihood using the Python-package PyDDM69.

The decision dynamics were described by (Figure 3A): Embedded Image where y is the decision variable, s is the stimulus category (−1 for left signals; 1 for right signals), v is the drift rate and controls the overall efficiency of evidence accumulation, vbias is the drift bias which is and evidence-independent constant added to the drift, and cdW is Gaussian distributed white noise with mean 0 and variance c2Δt. The starting point of evidence accumulation z was defined as a fraction of the boundary separation: Embedded Image where a is the separation between the two decision bounds. Evidence accumulation terminated at 0 (“left”) or a (“right). The model was fitted separately for each recording session of each participant. In each fit, we let drift rate vary with signal strength39, and starting point and drift bias with previous choice4.

Statistical comparisons

We used a paired sample t-test to test for significant differences in heart rate, pupil size, and behavioral measures between placebo and drug sessions (Figure 1 and Figure 3). We used 2 × 2 repeated measures ANOVA to test for the main effects of drug and signal strength, as well as their interaction, on behavioral measures (Figure 2). All tests were performed two-tailed.

We performed permutation analyses to obtain the distribution of correlation coefficients predicted exclusively by regression to the mean (Figure 2E). This was done by computing the above-described across-subject correlation 10K, each time using randomly assigned “placebo” vs. “drug” labels of each participant. We then compared our observed correlation coefficient (reflecting the combined effects of regression to the mean and the relationship between overall bias and the pupil predicted shift therein) against this permutation distribution. This analysis suggested that the relationship between choice history bias and the Atomoxetine predicted shift therein was stronger than expected based on regression to the mean (proportion of permutations below observed correlation < 0.05).

We also used a permutation procedure to test for differences between the correlation coefficients for history shifts in starting point bias and drift bias (Figure 3C). Again, we permuted the condition labels (drug / placebo).

Data availability

Data will be made publicly available upon publication.

Code availability

Analysis scripts will be made publicly available upon publication.

Supplementary figures

Figure S1.
Figure S1. Schematic of stimulus-response mapping rules.
Figure S2.
Figure S2. Atomoxetine does not affect overall choice bias.

(A) Signal detection theoretic sensitivity (Methods), separately for placebo and atomoxetine sessions, and separately or weak and strong signals. Every connecting line is a participant; large data points in the middle are the group averages (error bars, s.e.m. across 19 participants); stats, main effect of drug in two-way repeated measures ANOVA. (B) Histograms of reaction times, separately for placebo and atomoxetine sessions. (C) Permutation (randomized) distribution of correlation coefficients describing relationship between individual shift in repetition probability caused by atomoxetine and individual’s repetition probability during the placebo sessions. The observed correlation (plotted in Figure 2E) is stronger than 99% of this permutation distribution, resulting in a p-value of 0.01. (D) As A, but for signal detection theoretic criterion (Methods). (E) Individual shift in criterion caused by atomoxetine, plotted against individual’s criterion during the placebo sessions. Data points are individual participants; error bars, 68% confidence intervals across 5K bootstraps; stats, Pearson’s correlation coefficient (corrected for reversion to the mean; Methods). (F) As A, but for absolute criterion (Methods).

Figure S3.
Figure S3. Individual choice history biases are better explained by history-dependent changes in drift bias than starting point bias.

(A) Relationship between individual choice repetition probabilities [P(repeat)] and history shift in starting point (left column) and drift bias (right column), in the placebo sessions. Parameter estimates were obtained from a model in which both bias terms were allowed to vary with previous choice (Methods). Stats, Pearson’s correlation coefficient; black diamonds, outliers. (B) As A, but for atomoxetine sessions. (C) Group-level boxplots of all parameter estimates. (D) Measured and predicted RT distributions, separately for ‘left’ and ‘right’ choices, across all participants. (E) Conditional response bias functions. We divided all measured and model-predicted trials into five quantiles of the RT distribution and plotted for each quantile the fraction of ‘right’-choices. Error bars, s.e.m. across 19 participants.

Acknowledgments

Funding: Project grant of Amsterdam Brain & Cognition Priority Program; German Research Foundation (DFG; grant number Heisenberg Professorship DO 1240/3-1, and SFB 936, Projects A2/A3, A7, Z3).

Footnotes

  • Competing Interest Statement: The authors declare no competing interests.

References

  1. 1.↵
    Fischer, J. & Whitney, D. Serial dependence in visual perception. Nat Neurosci 17, 738–743 (2014).
    OpenUrlCrossRefPubMed
  2. 2.↵
    Tversky, A. & Kahneman, D. Judgment under Uncertainty: Heuristics and Biases. Science (New York, N.Y.) 185, 1124–1131 (1974).
    OpenUrlAbstract/FREE Full Text
  3. 3.↵
    Kloosterman, N. A. et al. Humans strategically shift decision bias by flexibly adjusting sensory evidence accumulation. eLife 8, e37321 (2019).
    OpenUrlCrossRefPubMed
  4. 4.↵
    Urai, A. E., de Gee, J. W., Tsetsos, K. & Donner, T. H. Choice history biases subsequent evidence accumulation. eLife 8, e46331 (2019).
    OpenUrlCrossRefPubMed
  5. 5.↵
    Braun, A., Urai, A. E. & Donner, T. H. Adaptive History Biases Result from Confidence-Weighted Accumulation of past Choices. J. Neurosci. 38, 2418–2429 (2018).
    OpenUrlAbstract/FREE Full Text
  6. 6.
    Hermoso-Mendizabal, A. et al. Response outcomes gate the impact of expectations on perceptual decisions. Nat Commun 11, 1057 (2020).
    OpenUrlCrossRefPubMed
  7. 7.
    Abrahamyan, A., Silva, L. L., Dakin, S. C., Carandini, M. & Gardner, J. L. Adaptable history biases in human perceptual decisions. PNAS 113, E3548–E3557 (2016).
    OpenUrlAbstract/FREE Full Text
  8. 8.↵
    Urai, A. E., Braun, A. & Donner, T. H. Pupil-linked arousal is driven by decision uncertainty and alters serial choice bias. Nat Commun 8, 14637 (2017).
    OpenUrlCrossRefPubMed
  9. 9.↵
    Urai, A. E. & Donner, T. H. Persistent activity in human parietal cortex mediates perceptual choice repetition bias. Nat Commun 13, 6015 (2022).
    OpenUrlCrossRefPubMed
  10. 10.
    Gold, J. I., Law, C.-T., Connolly, P. & Bennur, S. The Relative Influences of Priors and Sensory Evidence on an Oculomotor Decision Variable During Perceptual Learning. Journal of Neurophysiology 100, 2653–2668 (2008).
    OpenUrlCrossRefPubMedWeb of Science
  11. 11.
    Busse, L. et al. The Detection of Visual Contrast in the Behaving Mouse. J. Neurosci. 31, 11351–11361 (2011).
    OpenUrlAbstract/FREE Full Text
  12. 12.↵
    Braun, A. & Donner, T. H. Adaptive biasing of action-selective cortical build-up activity by stimulus history. Elife 12, RP86740 (2023).
    OpenUrlCrossRefPubMed
  13. 13.↵
    Fritsche, M. et al. Temporal regularities shape perceptual decisions and striatal dopamine signals. Nat Commun 15, 7093 (2024).
    OpenUrlCrossRefPubMed
  14. 14.↵
    1. Koller, D.,
    2. Schuurmans, D.,
    3. Bengio, Y. &
    4. Bottou, L.
    Yu, A. J. & Cohen, J. D. Sequential effects: Superstition or rational behavior? in Advances in Neural Information Processing Systems 21 (eds. Koller, D., Schuurmans, D., Bengio, Y. & Bottou, L.) 1873–1880 (Curran Associates, Inc., 2009).
  15. 15.↵
    Glaze, C. M., Kable, J. W. & Gold, J. I. Normative evidence accumulation in unpredictable environments. eLife 4, e08825 (2015).
    OpenUrlCrossRefPubMed
  16. 16.↵
    Mochol, G., Kiani, R. & Moreno-Bote, R. Prefrontal cortex represents heuristics that shape choice bias and its integration into future behavior. Current Biology 31, 1234-1244.e6 (2021).
    OpenUrlCrossRefPubMed
  17. 17.
    Akrami, A., Kopec, C. D., Diamond, M. E. & Brody, C. D. Posterior parietal cortex represents sensory history and mediates its effects on behaviour. Nature 554, 368–372 (2018).
    OpenUrlCrossRefPubMed
  18. 18.
    Hwang, E. J., Dahlen, J. E., Mukundan, M. & Komiyama, T. History-based action selection bias in posterior parietal cortex. Nat Commun 8, 1242 (2017).
    OpenUrlCrossRefPubMed
  19. 19.
    Morcos, A. S. & Harvey, C. D. History-dependent variability in population dynamics during evidence accumulation in cortex. Nat Neurosci 19, 1672–1681 (2016).
    OpenUrlCrossRefPubMed
  20. 20.↵
    Scott, B. B. et al. Fronto-parietal Cortical Circuits Encode Accumulated Evidence with a Diversity of Timescales. Neuron 95, 385-398.e5 (2017).
    OpenUrlCrossRefPubMed
  21. 21.↵
    Dayan, P. & Yu, A. J. Phasic norepinephrine: a neural interrupt signal for unexpected events. Network (Bristol, England) 17, 335–350 (2006).
    OpenUrlCrossRefPubMedWeb of Science
  22. 22.↵
    Yu, A. J. & Dayan, P. Uncertainty, neuromodulation, and attention. Neuron 46, 681–692 (2005).
    OpenUrlCrossRefPubMedWeb of Science
  23. 23.↵
    Aston-Jones, G. & Cohen, J. D. An Integrative Theory of Locus Coeruleus-Norepinephrine Function: Adaptive Gain and Optimal Performance. Annu. Rev. Neurosci. 28, 403–450 (2005).
    OpenUrlCrossRefPubMedWeb of Science
  24. 24.
    Harris, K. D. & Thiele, A. Cortical state and attention. Nature Reviews Neuroscience 12, 509–523 (2011).
    OpenUrlCrossRefPubMed
  25. 25.
    Lee, S.-H. & Dan, Y. Neuromodulation of brain states. Neuron 76, 209–222 (2012).
    OpenUrlCrossRefPubMedWeb of Science
  26. 26.
    McCormick, D. A., Nestvogel, D. B. & He, B. J. Neuromodulation of Brain State and Behavior. Annu Rev Neurosci 43, 391–415 (2020).
    OpenUrlCrossRefPubMed
  27. 27.
    Pfeffer, T. et al. Circuit mechanisms for the chemical modulation of cortex-wide network interactions and behavioral variability. Science Advances 7, eabf5620 (2021).
    OpenUrlFREE Full Text
  28. 28.↵
    Froemke, R. C. Plasticity of Cortical Excitatory-Inhibitory Balance. Annu. Rev. Neurosci. 38, 195–219 (2015).
    OpenUrlCrossRefPubMed
  29. 29.↵
    Joshi, S. & Gold, J. I. Pupil Size as a Window on Neural Substrates of Cognition. Trends in Cognitive Sciences (2020) doi:10.1016/j.tics.2020.03.005.
    OpenUrlCrossRefPubMed
  30. 30.↵
    Reimer, J. et al. Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Nat Commun 7, 13289 (2016).
    OpenUrlCrossRefPubMed
  31. 31.↵
    de Gee, J. W. et al. Dynamic modulation of decision biases by brainstem arousal systems. eLife 6, 309 (2017).
    OpenUrlCrossRef
  32. 32.↵
    Breton-Provencher, V., Drummond, G. T., Feng, J., Li, Y. & Sur, M. Spatiotemporal dynamics of noradrenaline during learned behaviour. Nature 606, 732–738 (2022).
    OpenUrlCrossRefPubMed
  33. 33.↵
    Bymaster, F. P. et al. Atomoxetine Increases Extracellular Levels of Norepinephrine and Dopamine in Prefrontal Cortex of Rat: A Potential Mechanism for Efficacy in Attention Deficit/Hyperactivity Disorder. Neuropsychopharmacology 27, 699–711 (2002).
    OpenUrlCrossRefPubMedWeb of Science
  34. 34.↵
    Robbins, T. W. & Arnsten, A. F. T. The Neuropsychopharmacology of Fronto-Executive Function: Monoaminergic Modulation. Annu. Rev. Neurosci. 32, 267–287 (2009).
    OpenUrlCrossRefPubMedWeb of Science
  35. 35.↵
    Steinmetz, N. A., Zatka-Haas, P., Carandini, M. & Harris, K. D. Distributed coding of choice, action and engagement across the mouse brain. Nature 576, 266–273 (2019).
    OpenUrlCrossRefPubMed
  36. 36.↵
    Gold, J. I. & Shadlen, M. N. The neural basis of decision making. Annu. Rev. Neurosci. 30, 535–574 (2007).
    OpenUrlCrossRefPubMedWeb of Science
  37. 37.↵
    Bogacz, R., Brown, E., Moehlis, J., Holmes, P. & Cohen, J. D. The physics of optimal decision making: a formal analysis of models of performance in two-alternative forced-choice tasks. Psychological review 113, 700–765 (2006).
    OpenUrlCrossRefPubMedWeb of Science
  38. 38.
    Brody, C. D. & Hanks, T. D. Neural underpinnings of the evidence accumulator. Current opinion in neurobiology 37, 149–157 (2016).
    OpenUrlCrossRefPubMed
  39. 39.↵
    Ratcliff, R. & McKoon, G. The diffusion decision model: theory and data for two-choice decision tasks. Neural computation 20, 873–922 (2008).
    OpenUrlCrossRefPubMedWeb of Science
  40. 40.↵
    Cremer, A., Kalbe, F., Müller, J. C., Wiedemann, K. & Schwabe, L. Disentangling the roles of dopamine and noradrenaline in the exploration-exploitation tradeoff during human decision-making. Neuropsychopharmacol. 1–9 (2022) doi:10.1038/s41386-022-01517-9.
    OpenUrlCrossRef
  41. 41.
    Jepma, M. et al. Noradrenergic and Cholinergic Modulation of Belief Updating. Journal of Cognitive Neuroscience 30, 1803–1820 (2018).
    OpenUrlCrossRefPubMed
  42. 42.
    Lawson, R. P., Bisby, J., Nord, C. L., Burgess, N. & Rees, G. The Computational, Pharmacological, and Physiological Determinants of Sensory Learning under Uncertainty. Current Biology 31, 163-172.e4 (2021).
    OpenUrlCrossRefPubMed
  43. 43.↵
    Reynaud, A. J. et al. Atomoxetine improves attentional orienting in a predictive context. Neuropharmacology 150, 59–69 (2019).
    OpenUrlCrossRefPubMed
  44. 44.↵
    Totah, N. K., Logothetis, N. K. & Eschenko, O. Atomoxetine accelerates attentional set shifting without affecting learning rate in the rat. Psychopharmacology 232, 3697–3707 (2015).
    OpenUrlCrossRef
  45. 45.↵
    Filipowicz, A. L., Glaze, C. M., Kable, J. W. & Gold, J. I. Pupil diameter encodes the idiosyncratic, cognitive complexity of belief updating. Elife 9, e57872 (2020).
    OpenUrlCrossRefPubMed
  46. 46.
    Findling, C., Skvortsova, V., Dromnelle, R., Palminteri, S. & Wyart, V. Computational noise in reward-guided learning drives behavioral variability in volatile environments. Nat Neurosci 22, 2066–2077 (2019).
    OpenUrlCrossRefPubMed
  47. 47.
    Krishnamurthy, K., Nassar, M. R., Sarode, S. & Gold, J. I. Arousal-related adjustments of perceptual biases optimize perception in dynamic environments. Nature Human Behaviour 1, 0107 (2017).
    OpenUrlCrossRefPubMed
  48. 48.
    Murphy, P. R., Wilming, N., Hernandez-Bocanegra, D. C., Prat-Ortega, G. & Donner, T. H. Adaptive circuit dynamics across human cortex during evidence accumulation in changing environments. Nat Neurosci 24, 987–997 (2021).
    OpenUrlCrossRefPubMed
  49. 49.↵
    Nassar, M. R. et al. Rational regulation of learning dynamics by pupil-linked arousal systems. Nat Neurosci 15, 1040–1046 (2012).
    OpenUrlCrossRefPubMed
  50. 50.↵
    Breton-Provencher, V. & Sur, M. Active control of arousal by a locus coeruleus GABAergic circuit. Nat Neurosci 22, 218–228 (2019).
    OpenUrlCrossRefPubMed
  51. 51.
    Joshi, S., Li, Y., Kalwani, R. M. & Gold, J. I. Relationships between Pupil Diameter and Neuronal Activity in the Locus Coeruleus, Colliculi, and Cingulate Cortex. Neuron 89, 221–234 (2016).
    OpenUrlCrossRefPubMed
  52. 52.
    Murphy, P. R., O’Connell, R. G., O’Sullivan, M., Robertson, I. H. & Balsters, J. H. Pupil diameter covaries with BOLD activity in human locus coeruleus. Human brain mapping 35, 4140–4154 (2014).
    OpenUrlCrossRefPubMedWeb of Science
  53. 53.↵
    Varazzani, C., San-Galli, A., Gilardeau, S. & Bouret, S. Noradrenaline and Dopamine Neurons in the Reward/Effort Trade-Off: A Direct Electrophysiological Comparison in Behaving Monkeys. J. Neurosci. 35, 7866–7877 (2015).
    OpenUrlAbstract/FREE Full Text
  54. 54.↵
    Mridha, Z. et al. Graded recruitment of pupil-linked neuromodulation by parametric stimulation of the vagus nerve. Nat Commun 12, 1539 (2021).
    OpenUrlCrossRefPubMed
  55. 55.↵
    de Gee, J. W. et al. Pupil-linked phasic arousal predicts a reduction of choice bias across species and decision domains. Elife 9, e54014 (2020).
    OpenUrlCrossRefPubMed
  56. 56.
    Lewandowska, K., Gągol, A., Sikora-Wachowicz, B., Marek, T. & Fąfrowicz, M. Saying “yes” when you want to say “no” - pupil dilation reflects evidence accumulation in a visual working memory recognition task. International Journal of Psychophysiology 139, 18–32 (2019).
    OpenUrlCrossRefPubMed
  57. 57.
    Schriver, B. J., Perkins, S. M., Sajda, P. & Wang, Q. Interplay between components of pupil-linked phasic arousal and its role in driving behavioral choice in Go/No-Go perceptual decision-making. Psychophysiology 00, e13565 (2020).
    OpenUrl
  58. 58.↵
    de Gee, J. W., Knapen, T. & Donner, T. H. Decision-related pupil dilation reflects upcoming choice and individual bias. Proceedings of the National Academy of Sciences of the United States of America 111, E618–25 (2014).
    OpenUrlAbstract/FREE Full Text
  59. 59.↵
    Bari, A. & Aston-Jones, G. Atomoxetine modulates spontaneous and sensory-evoked discharge of locus coeruleus noradrenergic neurons. Neuropharmacology 64, 53–64 (2013).
    OpenUrlCrossRefPubMedWeb of Science
  60. 60.↵
    Cazettes, F., Reato, D., Morais, J. P., Renart, A. & Mainen, Z. F. Phasic Activation of Dorsal Raphe Serotonergic Neurons Increases Pupil Size. Current Biology 31, 192-197.e4 (2021).
    OpenUrlCrossRefPubMed
  61. 61.↵
    Grujic, N., Tesmer, A., Bracey, E., Peleg-Raibstein, D. & Burdakov, D. Control and coding of pupil size by hypothalamic orexin neurons. Nat Neurosci 26, 1160–1164 (2023).
    OpenUrlCrossRefPubMed
  62. 62.↵
    Sara, S. J. The locus coeruleus and noradrenergic modulation of cognition. Nature Reviews Neuroscience 10, 211–223 (2009).
    OpenUrlCrossRefPubMedWeb of Science
  63. 63.↵
    Ranjbar-Slamloo, Y. & Fazlali, Z. Dopamine and Noradrenaline in the Brain; Overlapping or Dissociate Functions? Front. Mol. Neurosci. 12, (2020).
  64. 64.↵
    Ding, Y.-S. et al. Clinical doses of atomoxetine significantly occupy both norepinephrine and serotonin transports: Implications on treatment of depression and ADHD. NeuroImage 86, 164–171 (2014).
    OpenUrlCrossRefPubMedWeb of Science
  65. 65.↵
    Sauer, J.-M., Ring, B. J. & Witcher, J. W. Clinical Pharmacokinetics of Atomoxetine. Clin Pharmacokinet 44, 571–590 (2005).
    OpenUrlCrossRefPubMedWeb of Science
  66. 66.↵
    Watson, A. B. & Pelli, D. G. QUEST: A Bayesian adaptive psychometric method. Perception & psychophysics (1983).
  67. 67.↵
    Knapen, T. et al. Cognitive and Ocular Factors Jointly Determine Pupil Responses under Equiluminance. PLOS ONE 11, e0155574 (2016).
    OpenUrlCrossRefPubMed
  68. 68.↵
    Green, D. M. & Swets, J. A. Signal Detection Theory and Psychophysics. 1966. (New York, 1966).
  69. 69.↵
    Shinn, M., Lam, N. H. & Murray, J. D. A flexible framework for simulating and fitting generalized drift-diffusion models. eLife 9, e56938.
Back to top
Posted October 28, 2024.
Download PDF
Email
Catecholamines reduce choice history biases
Jan Willem de Gee, Niels A. Kloosterman, Anke Braun, Tobias H. Donner
bioRxiv 2024.10.28.620689; doi: https://doi.org/10.1101/2024.10.28.620689
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area