2025
Fleming, Hugo; Feng, Gloria W; Rutledge, Robb B; Roiser, Jonathan P; Robinson, Oliver J
Cognitive training reduces the strength of Pavlovian biases. Journal Article
In: Journal of Experimental Psychology: Learning, Memory, and Cognition, 2025.
@article{fleming2025cognitive,
title = {Cognitive training reduces the strength of Pavlovian biases.},
author = {Hugo Fleming and Gloria W Feng and Robb B Rutledge and Jonathan P Roiser and Oliver J Robinson},
doi = {https://doi.org/10.1037/xlm0001561},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
journal = {Journal of Experimental Psychology: Learning, Memory, and Cognition},
publisher = {American Psychological Association},
abstract = {Pavlovian biases are patterns of behavior that involve approaching stimuli associated with reward and avoiding those associated with punishment (regardless of whether this is actually optimal behavior). They are a ubiquitous feature of everyday decision making and are also believed to play an important role in the symptoms of anxiety and depression. Although Pavlovian biases have classically been described as fixed and automatic, some studies have indicated that their influence on behavior can actually vary over time and with task demands. While these results hint that people may have some control over their Pavlovian biases, direct behavioral evidence for this control is still lacking. In a preregistered, double-blind, sham-controlled study (N = 800), we tested whether a week-long cognitive training intervention could reduce Pavlovian biases on the orthogonalized go/no-go task, a well-established paradigm for isolating Pavlovian-instrumental conflict. Participants were trained on either high-conflict or no-conflict conditions of the task across 5 days. Using reinforcement learning models to dissociate components of decision making, we found that high-conflict training led to a significant reduction in Pavlovian bias—particularly avoidance bias—at follow-up. This result is incompatible with the view that Pavlovian biases are fixed and automatic, and instead implies much greater flexibility in the way that they influence cognition than has previously been understood. The training was kept deliberately simple (i.e., one stimulus per condition, with the correct responses kept constant over sessions) so as to provide a minimal proof of concept of whether Pavlovian biases can be reduced through training, but as a result, we did not observe transfer to other tasks or self-reported mood. Nonetheless, these findings demonstrate that targeted cognitive training can modulate Pavlovian biases, which may be beneficial both in everyday life and especially in the context of affective disorders like anxiety and depression.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Pavlovian biases are patterns of behavior that involve approaching stimuli associated with reward and avoiding those associated with punishment (regardless of whether this is actually optimal behavior). They are a ubiquitous feature of everyday decision making and are also believed to play an important role in the symptoms of anxiety and depression. Although Pavlovian biases have classically been described as fixed and automatic, some studies have indicated that their influence on behavior can actually vary over time and with task demands. While these results hint that people may have some control over their Pavlovian biases, direct behavioral evidence for this control is still lacking. In a preregistered, double-blind, sham-controlled study (N = 800), we tested whether a week-long cognitive training intervention could reduce Pavlovian biases on the orthogonalized go/no-go task, a well-established paradigm for isolating Pavlovian-instrumental conflict. Participants were trained on either high-conflict or no-conflict conditions of the task across 5 days. Using reinforcement learning models to dissociate components of decision making, we found that high-conflict training led to a significant reduction in Pavlovian bias—particularly avoidance bias—at follow-up. This result is incompatible with the view that Pavlovian biases are fixed and automatic, and instead implies much greater flexibility in the way that they influence cognition than has previously been understood. The training was kept deliberately simple (i.e., one stimulus per condition, with the correct responses kept constant over sessions) so as to provide a minimal proof of concept of whether Pavlovian biases can be reduced through training, but as a result, we did not observe transfer to other tasks or self-reported mood. Nonetheless, these findings demonstrate that targeted cognitive training can modulate Pavlovian biases, which may be beneficial both in everyday life and especially in the context of affective disorders like anxiety and depression.
Brener, S; Feng, G; Merrick, C; Blain, B; Rutledge, R; Little, S
Theta and Beta Oscillations Track Happiness and Surprise in the Subthalamic Nucleus After Reward Outcome Proceedings Article
In: MOVEMENT DISORDERS, WILEY 111 RIVER ST, HOBOKEN 07030-5774, NJ USA 2025.
@inproceedings{brener2025theta,
title = {Theta and Beta Oscillations Track Happiness and Surprise in the Subthalamic Nucleus After Reward Outcome},
author = {S Brener and G Feng and C Merrick and B Blain and R Rutledge and S Little},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
booktitle = {MOVEMENT DISORDERS},
volume = {40},
organization = {WILEY 111 RIVER ST, HOBOKEN 07030-5774, NJ USA},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
2024
Feng, Gloria W; Rutledge, Robb B
Surprising sounds influence risky decision making Journal Article
In: Nature Communications, vol. 15, no. 1, pp. 8027, 2024.
@article{Feng2024,
title = {Surprising sounds influence risky decision making},
author = {Gloria W Feng and Robb B Rutledge},
doi = {https://doi.org/10.1038/s41467-024-51729-4},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Nature Communications},
volume = {15},
number = {1},
pages = {8027},
publisher = {Nature Publishing Group UK London},
abstract = {Adaptive behavior depends on appropriate responses to environmental uncertainty. Incidental sensory events might simply be distracting and increase errors, but alternatively can lead to stereotyped responses despite their irrelevance. To evaluate these possibilities, we test whether task-irrelevant sensory prediction errors influence risky decision making in humans across seven experiments (total n = 1600). Rare auditory sequences preceding option presentation systematically increase risk taking and decrease choice perseveration (i.e., increased tendency to switch away from previously chosen options). The risk-taking and perseveration effects are dissociable by manipulating auditory statistics: when rare sequences end on standard tones, including when rare sequences consist only of standard tones, participants are less likely to perseverate after rare sequences but not more likely to take risks. Computational modeling reveals that these effects cannot be explained by increased decision noise but can be explained by value-independent risky bias and perseveration parameters, decision biases previously linked to dopamine. Control experiments demonstrate that both surprise effects can be eliminated when tone sequences are presented in a balanced or fully predictable manner, and that surprise effects cannot be explained by erroneous beliefs. These findings suggest that incidental sounds may influence many of the decisions we make in daily life.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Adaptive behavior depends on appropriate responses to environmental uncertainty. Incidental sensory events might simply be distracting and increase errors, but alternatively can lead to stereotyped responses despite their irrelevance. To evaluate these possibilities, we test whether task-irrelevant sensory prediction errors influence risky decision making in humans across seven experiments (total n = 1600). Rare auditory sequences preceding option presentation systematically increase risk taking and decrease choice perseveration (i.e., increased tendency to switch away from previously chosen options). The risk-taking and perseveration effects are dissociable by manipulating auditory statistics: when rare sequences end on standard tones, including when rare sequences consist only of standard tones, participants are less likely to perseverate after rare sequences but not more likely to take risks. Computational modeling reveals that these effects cannot be explained by increased decision noise but can be explained by value-independent risky bias and perseveration parameters, decision biases previously linked to dopamine. Control experiments demonstrate that both surprise effects can be eliminated when tone sequences are presented in a balanced or fully predictable manner, and that surprise effects cannot be explained by erroneous beliefs. These findings suggest that incidental sounds may influence many of the decisions we make in daily life.
Hur, Jihyun K; Heffner, Joseph; Feng, Gloria W; Joormann, Jutta; Rutledge, Robb B
Language sentiment predicts changes in depressive symptoms Journal Article
In: Proceedings of the National Academy of Sciences, vol. 121, no. 39, pp. e2321321121, 2024.
@article{hur2024language,
title = {Language sentiment predicts changes in depressive symptoms},
author = {Jihyun K Hur and Joseph Heffner and Gloria W Feng and Jutta Joormann and Robb B Rutledge},
doi = {https://doi.org/10.1073/pnas.2321321121},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Proceedings of the National Academy of Sciences},
volume = {121},
number = {39},
pages = {e2321321121},
publisher = {National Academy of Sciences},
abstract = {The prevalence of depression is a major societal health concern, and there is an ongoing need to develop tools that predict who will become depressed. Past research suggests that depression changes the language we use, but it is unclear whether language is predictive of worsening symptoms. Here, we test whether the sentiment of brief written linguistic responses predicts changes in depression. Across two studies (N = 467), participants provided responses to neutral open-ended questions, narrating aspects of their lives relevant to depression (e.g., mood, motivation, sleep). Participants also completed the Patient Health Questionnaire (PHQ-9) to assess depressive symptoms and a risky decision-making task with periodic measurements of momentary happiness to quantify mood dynamics. The sentiment of written responses was evaluated by human raters (N = 470), Large Language Models (LLMs; ChatGPT 3.5 and 4.0), and the Linguistic Inquiry and Word Count (LIWC) tool. We found that language sentiment evaluated by human raters and LLMs, but not LIWC, predicted changes in depressive symptoms at a three-week follow-up. Using computational modeling, we found that language sentiment was associated with current mood, but language sentiment predicted symptom changes even after controlling for current mood. In summary, we demonstrate a scalable tool that combines brief written responses with sentiment analysis by AI tools that matches human performance in the prediction of future psychiatric symptoms.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The prevalence of depression is a major societal health concern, and there is an ongoing need to develop tools that predict who will become depressed. Past research suggests that depression changes the language we use, but it is unclear whether language is predictive of worsening symptoms. Here, we test whether the sentiment of brief written linguistic responses predicts changes in depression. Across two studies (N = 467), participants provided responses to neutral open-ended questions, narrating aspects of their lives relevant to depression (e.g., mood, motivation, sleep). Participants also completed the Patient Health Questionnaire (PHQ-9) to assess depressive symptoms and a risky decision-making task with periodic measurements of momentary happiness to quantify mood dynamics. The sentiment of written responses was evaluated by human raters (N = 470), Large Language Models (LLMs; ChatGPT 3.5 and 4.0), and the Linguistic Inquiry and Word Count (LIWC) tool. We found that language sentiment evaluated by human raters and LLMs, but not LIWC, predicted changes in depressive symptoms at a three-week follow-up. Using computational modeling, we found that language sentiment was associated with current mood, but language sentiment predicted symptom changes even after controlling for current mood. In summary, we demonstrate a scalable tool that combines brief written responses with sentiment analysis by AI tools that matches human performance in the prediction of future psychiatric symptoms.
2023
Kao, Chang-Hao; Feng, Gloria W; Hur, Jihyun K; Jarvis, Huw; Rutledge, Robb B
Computational models of subjective feelings in psychiatry Journal Article
In: Neuroscience & Biobehavioral Reviews, vol. 145, pp. 105008, 2023.
@article{kao2023computational,
title = {Computational models of subjective feelings in psychiatry},
author = {Chang-Hao Kao and Gloria W Feng and Jihyun K Hur and Huw Jarvis and Robb B Rutledge},
doi = {https://doi.org/10.1016/j.neubiorev.2022.105008},
year = {2023},
date = {2023-01-01},
urldate = {2023-01-01},
journal = {Neuroscience & Biobehavioral Reviews},
volume = {145},
pages = {105008},
publisher = {Elsevier},
abstract = {Research in computational psychiatry is dominated by models of behavior. Subjective experience during behavioral tasks is not well understood, even though it should be relevant to understanding the symptoms of psychiatric disorders. Here, we bridge this gap and review recent progress in computational models for subjective feelings. For example, happiness reflects not how well people are doing, but whether they are doing better than expected. This dependence on recent reward prediction errors is intact in major depression, although depressive symptoms lower happiness during tasks. Uncertainty predicts subjective feelings of stress in volatile environments. Social prediction errors influence feelings of self-worth more in individuals with low self-esteem despite a reduced willingness to change beliefs due to social feedback. Measuring affective state during behavioral tasks provides a tool for understanding psychiatric symptoms that can be dissociable from behavior. When smartphone tasks are collected longitudinally, subjective feelings provide a potential means to bridge the gap between lab-based behavioral tasks and real-life behavior, emotion, and psychiatric symptoms.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Research in computational psychiatry is dominated by models of behavior. Subjective experience during behavioral tasks is not well understood, even though it should be relevant to understanding the symptoms of psychiatric disorders. Here, we bridge this gap and review recent progress in computational models for subjective feelings. For example, happiness reflects not how well people are doing, but whether they are doing better than expected. This dependence on recent reward prediction errors is intact in major depression, although depressive symptoms lower happiness during tasks. Uncertainty predicts subjective feelings of stress in volatile environments. Social prediction errors influence feelings of self-worth more in individuals with low self-esteem despite a reduced willingness to change beliefs due to social feedback. Measuring affective state during behavioral tasks provides a tool for understanding psychiatric symptoms that can be dissociable from behavior. When smartphone tasks are collected longitudinally, subjective feelings provide a potential means to bridge the gap between lab-based behavioral tasks and real-life behavior, emotion, and psychiatric symptoms.
Fleming, Hugo Alexander; Feng, Gloria; Rutledge, Robb; Roiser, Jonathan; Robinson, Oliver J
Training successfully reduces the strength of Pavlovian biases Journal Article
In: 2023.
@article{fleming2023training,
title = {Training successfully reduces the strength of Pavlovian biases},
author = {Hugo Alexander Fleming and Gloria Feng and Robb Rutledge and Jonathan Roiser and Oliver J Robinson},
doi = {https://doi.org/10.31234/osf.io/h3297},
year = {2023},
date = {2023-01-01},
urldate = {2023-01-01},
publisher = {PsyArXiv},
abstract = {Pavlovian biases are fixed patterns of responses that include approaching stimuli associated with reward and avoiding those associated with punishment. These prepotent behavioural responses can sometimes conflict with those produced by other behavioural action selection systems, giving rise to suboptimal behaviour. This is particularly important in the context of affective disorders like anxiety and depression, in which Pavlovian biases are enhanced (Mkrtchian et al., 2017; Nord et al., 2018), and are thought to contribute to the induction and maintenance of symptoms (Dayan & Huys, 2008). In this study we investigated whether participants could be trained to exert more control over these biases. This would present a potential new opportunity for treating anxiety and depression. In addition, it would be the first direct, behavioural demonstration that Pavlovian biases are modifiable. We conducted a double-blind, sham-controlled study (N= 800) and found that the active training intervention was indeed effective: after selectively practicing the conflict trials of the Orthogonal Go/No-Go task (Guitart-Masip et al., 2011), which require high control, participants were more accurate and showed less influence of Pavlovian biases (especially avoidance bias) when tested on the full task. We discuss what this means for understanding Pavlovian biases, and suggest that future studies should now be aimed at facilitating transfer of these effects to other aspects of anxiety and depression symptoms.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Pavlovian biases are fixed patterns of responses that include approaching stimuli associated with reward and avoiding those associated with punishment. These prepotent behavioural responses can sometimes conflict with those produced by other behavioural action selection systems, giving rise to suboptimal behaviour. This is particularly important in the context of affective disorders like anxiety and depression, in which Pavlovian biases are enhanced (Mkrtchian et al., 2017; Nord et al., 2018), and are thought to contribute to the induction and maintenance of symptoms (Dayan & Huys, 2008). In this study we investigated whether participants could be trained to exert more control over these biases. This would present a potential new opportunity for treating anxiety and depression. In addition, it would be the first direct, behavioural demonstration that Pavlovian biases are modifiable. We conducted a double-blind, sham-controlled study (N= 800) and found that the active training intervention was indeed effective: after selectively practicing the conflict trials of the Orthogonal Go/No-Go task (Guitart-Masip et al., 2011), which require high control, participants were more accurate and showed less influence of Pavlovian biases (especially avoidance bias) when tested on the full task. We discuss what this means for understanding Pavlovian biases, and suggest that future studies should now be aimed at facilitating transfer of these effects to other aspects of anxiety and depression symptoms.
2021
Globig, Laura K; Witte, Kristin; Feng, Gloria; Sharot, Tali
Under threat, weaker evidence is required to reach undesirable conclusions Journal Article
In: Journal of Neuroscience, vol. 41, no. 30, pp. 6502–6510, 2021.
@article{globig2021under,
title = {Under threat, weaker evidence is required to reach undesirable conclusions},
author = {Laura K Globig and Kristin Witte and Gloria Feng and Tali Sharot},
doi = {https://doi.org/10.1523/JNEUROSCI.3194-20.2021},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {Journal of Neuroscience},
volume = {41},
number = {30},
pages = {6502–6510},
publisher = {Soc Neuroscience},
abstract = {Critical decisions, such as in domains ranging from medicine to finance, are often made under threatening circumstances that elicit stress and anxiety. The negative effects of such reactions on learning and decision-making have been repeatedly underscored. In contrast, here we show that perceived threat alters the process by which evidence is accumulated in a way that may be adaptive. Participants (n = 91) completed a sequential evidence sampling task in which they were incentivized to accurately judge whether they were in a desirable state, which was associated with greater rewards than losses, or an undesirable state, which was associated with greater losses than rewards. Before the task participants in the “threat group” experienced a social-threat manipulation. Results show that perceived threat led to a reduction in the strength of evidence required to reach an undesirable judgment. Computational modeling revealed this was because of an increase in the relative rate by which negative information was accumulated. The effect of the threat manipulation was global, as the alteration to evidence accumulation was observed for information which was not directly related to the cause of the threat. Requiring weaker evidence to reach undesirable conclusions in threatening environments may be adaptive as it can lead to increased precautionary action.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Critical decisions, such as in domains ranging from medicine to finance, are often made under threatening circumstances that elicit stress and anxiety. The negative effects of such reactions on learning and decision-making have been repeatedly underscored. In contrast, here we show that perceived threat alters the process by which evidence is accumulated in a way that may be adaptive. Participants (n = 91) completed a sequential evidence sampling task in which they were incentivized to accurately judge whether they were in a desirable state, which was associated with greater rewards than losses, or an undesirable state, which was associated with greater losses than rewards. Before the task participants in the “threat group” experienced a social-threat manipulation. Results show that perceived threat led to a reduction in the strength of evidence required to reach an undesirable judgment. Computational modeling revealed this was because of an increase in the relative rate by which negative information was accumulated. The effect of the threat manipulation was global, as the alteration to evidence accumulation was observed for information which was not directly related to the cause of the threat. Requiring weaker evidence to reach undesirable conclusions in threatening environments may be adaptive as it can lead to increased precautionary action.