JLJessica C. Lee

Jessica C. Lee

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Jessica C. Lee has a total of 190 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Benchmarks for Associative Learning Models

    by Dominik R Bach, Ralph R. Miller, Jose E. Callejas-Aguilera, Hilary Don, Kurt Fraser, Mihaela Iordonova, Jessica C. Lee, Josie Linnell, Huaiyu Liu, Evan James Livesey, et al.

    Associative learning, like many mature fields of behavioural science, is endowed with a nearly unfathomable wealth of empirical data. These arise from multiple experimental procedures and animal species. To assess any proposed theory of associative learning, decisions must be made about which behavioral phenomena to account for. However, there is currently no systematic approach to selecting observations within this body of evidence. Here, we propose benchmarks for theories of associative learning by organizing phenomena according to their robustness across experiments, their generality across procedural variations, domains, and species, and the degree to which they occur under broad but clearly specified circumstances. We queried the expert community for phenomena to be included, reviewed the relevant literature, classified the phenomena into four categories of robustness and generality, and compared our classification with a second survey of the expert community. We identified 92 benchmark phenomena, of which 21 are domain- and species-general, and 24 are species- and/or domain-specific but highly robust across procedures. Strikingly, some textbook phenomena that have motivated influential and general associative learning theories are not among these. The remaining 47 phenomena were either constrained to particular experimental procedures or of inconclusive generality. Some of their implementational boundary conditions appear to be informally known but were rarely investigated systematically. In summary, our review provides a comprehensive evaluation of the current evidence, which will support theory building while also re-focusing research into areas in which evidence is lacking.

  • Open Access

    Is Generalization General? Differences Between Predictive and Category Learning Reflect Knowledge of “Cognitive Kinds”

    by Jessica C. Lee, Andy J. Wills, René Schlegelmilch

    Shepard (1987) observed that generalization gradients were invariant between species and stimulus dimensions, and proposed that these universal properties reflect knowledge of “natural kinds” that exist in the world. Here, we test a natural extension of Shepard’s ideas, by examining whether generalization differs between different types of learned associations. To do so, we leverage generalization as a tool to diagnose the underlying content of learning in two domains (predictive learning and category learning), while equating stimuli and procedural details between tasks. Experiment 1 showed that empirical differences in generalization gradients between domains can be largely explained by differences in the testing procedures. When these were equated, the underlying generalization function was largely similar between predictive and category learning. The results of Experiment 2 were more complex, revealing some differences between domains which may be explained by differences in the representation of outcomes and categories. We conclude that the similarities between learning domains (underlying gradient shape and response to experimental manipulations) provide support for Shepard’s (1987) proposal that generalization exhibits universal properties, and propose that the differences between learning domains reflect knowledge about “cognitive kinds”; categories are created to be mutually exclusive while outcomes are not.

  • Open Access

    Investigating the reproducibility of the social and behavioural sciences

    by Olivia Miske, Anna Lou Abatayo, Macie Daley, Mirka Dirzo, Nicholas William Fox, Noah Haber, Krystal Hahn, Melissa Kline Struhl, Brinna Mawhinney, Priya Silverstein, et al.

    Published claims should be reproducible, yielding the same result when applying the same analysis to the same data. We assessed reproducibility in a stratified random sample of 600 papers published from 2009 to 2018 in 62 journals spanning the social and behavioural sciences. Authors of 144 (24.0% [95% CI 20.8 - 27.6%]) papers made data available to assess reproducibility, and for 38 others, we obtained source data to reconstruct the dataset. We assessed 143 of the 182 available datasets and found that 76.6 (53.6% [95% CI 45.8 - 60.7%]) papers were rated as precisely reproducible and 105.0 (73.5% [95% CI 66.4 - 80.0%]) papers as at least approximately reproducible (within 15% of the original effects or within .05 of original p-values) after inverse weighting each of the 551 claims by the number of claims per paper. We observed higher reproducibility for papers from Political Science and Economics than other fields, for more recent than older papers, and for papers from journals that required data sharing. Implementing measures to verify that research is reproducible is needed to support trustworthiness in the complex enterprise of knowledge production.

  • Open Access

    Investigating the reproducibility of the social and behavioural sciences

    by Olivia Miske, Anna Lou Abatayo, Macie Daley, Mirka Dirzo, Nicholas William Fox, Noah Haber, Krystal Hahn, Melissa Kline Struhl, Brinna Mawhinney, Priya Silverstein, et al.

    Published claims should be reproducible, yielding the same result when applying the same analysis to the same data. We assessed reproducibility in a stratified random sample of 600 papers published from 2009 to 2018 in 62 journals spanning the social and behavioural sciences. Authors of 144 (24.0% [95% CI 20.8 - 27.6%]) papers made data available to assess reproducibility, and for 38 others, we obtained source data to reconstruct the dataset. We assessed 143 of the 182 available datasets and found that 76.6 (53.6% [95% CI 45.8 - 60.7%]) papers were rated as precisely reproducible and 105.0 (73.5% [95% CI 66.4 - 80.0%]) papers as at least approximately reproducible (within 15% of the original effects or within .05 of original p-values) after inverse weighting each of the 551 claims by the number of claims per paper. We observed higher reproducibility for papers from Political Science and Economics than other fields, for more recent than older papers, and for papers from journals that required data sharing. Implementing measures to verify that research is reproducible is needed to support trustworthiness in the complex enterprise of knowledge production.

  • Open Access

    Investigating the reproducibility of the social and behavioural sciences

    by Olivia Miske, Anna Lou Abatayo, Macie Daley, Mirka Dirzo, Nicholas William Fox, Noah Haber, Krystal Hahn, Melissa Kline Struhl, Brinna Mawhinney, Priya Silverstein, et al.

    Published claims should be reproducible, yielding the same result when applying the same analysis to the same data. We assessed reproducibility in a stratified random sample of 600 papers published from 2009 to 2018 in 62 journals spanning the social and behavioural sciences. Authors of 144 (24.0% [95% CI 20.8 - 27.6%]) papers made data available to assess reproducibility, and for 38 others, we obtained source data to reconstruct the dataset. We assessed 143 of the 182 available datasets and found that 76.6 (53.6% [95% CI 45.8 - 60.7%]) papers were rated as precisely reproducible and 105.0 (73.5% [95% CI 66.4 - 80.0%]) papers as at least approximately reproducible (within 15% of the original effects or within .05 of original p-values) after inverse weighting each of the 551 claims by the number of claims per paper. We observed higher reproducibility for papers from Political Science and Economics than other fields, for more recent than older papers, and for papers from journals that required data sharing. Implementing measures to verify that research is reproducible is needed to support trustworthiness in the complex enterprise of knowledge production.

  • Open Access

    When time matters: generalization gradients in delay and trace conditioning procedures

    by José A. Alcalá, Jessica C. Lee, Gonzalo Pablo Urcelay

    Previous studies have documented that overshadowing weakens when a CS-US trace is inserted, and we have interpreted such effect in terms of differential generalization in trace versus delay procedures, but direct evidence in humans is lacking. Two online experiments investigated whether the weakening of cue-outcome temporal contiguity impacts generalization gradients. Participants experienced strong temporal contiguity (the offset of a cue coincides with the onset an outcome) or weak temporal contiguity (there is a temporal gap between the offset of a cue and the onset of an outcome) in a continuous predictive learning task. Initial training involved a double negative discrimination, in which there was one reinforced cue (aqua-color rectangle) and two non-reinforced cues (one greener and one bluer than the reinforced cue), and participants had to make a response whenever they predicted the outcome. After training, in the generalization test phase, outcome expectancy in presence of several stimuli along the full green-blue dimension was tested. Outcome expectancy was assessed with discrete ratings in Experiment 1 (n=252) and a behavioral dependent measure (button presses, similar to the training phase) in Experiment 2 (n=250). Regardless the type of measure, when participants experienced weak cue-outcome temporal contiguity showed a broader generalization response, although differences were more prominent in the green side of the color dimension. Furthermore, no significant differences emerged between two groups experiencing strong temporal contiguity but differing in cue duration (controlling stimulus onset asynchrony, SOA), suggesting that cue-outcome temporal contiguity plays a more critical role than cue duration in shaping generalization. Overall, this result advances our understanding of the role of temporal contiguity at a basic level, while opening the possibility to study it as a critical factor in more applied settings.

  • Open Access

    Trait anxiety and fear generalization: Overgeneralization of fear or undergeneralization of safety learning?

    by Jessica C. Lee, Tracy Dang, Jack Collins

    The tendency to overgeneralize fear learning has been identified as a potential risk factor for anxiety disorders. In this study, we examined whether highly anxious individuals differed from low anxious individuals in how they generalized fear (learning when an aversive stimulus is present) and safety (learning when an aversive stimulus is absent) following differential conditioning with an aversive scream outcome. We achieved this by using a morphed shape dimension to separately measure generalization from the fear cue (predicting the scream) and generalization from the safety cue (predicting the absence of the scream). In two experiments, we found that relative to low trait anxious participants, high trait anxious participants showed higher outcome predictions to stimuli resembling the safety cue (i.e., they undergeneralized safety learning), but not for stimuli resembling the fear cue (i.e., they did not overgeneralize fear learning). Undergeneralization was not found when a neutral outcome was used, suggesting that this effect is dependent on the use of an aversive outcome and specific to safety learning. Our findings suggest that safety generalization may vary more as a function of trait anxiety than fear generalization, and therefore future research should separately measure these processes to uncover the mechanisms driving excessive spread of fear.

  • Open Access

    Illusory or normative? Selective inferences about hidden causes can protect illusory causal beliefs

    by Peter Lovibond, Julie Chow, Jessica C. Lee

    Our ability to extract causal information from direct experience is crucial in making sense of our world. It is therefore puzzling that people often overestimate the relationship between objectively unrelated events. Such causal illusions can be investigated in a contingency learning task, where participants judge the effectiveness of a cue in causing a target outcome based on their observed covariation across a series of trials. When cue and outcome co-occur (cell A trials), the outcome is readily attributed to the cue. But how do participants interpret outcomes in the absence of the cue (cell C trials), where the cause of the outcome is unspecified? We asked participants to provide open-ended (Experiment 1) or forced-choice (Experiments 2 and 3) explanations for trials in a causal illusion task where the cue was a medicine and the outcome was recovery from a disease. Participants inferred a range of plausible hidden causes for cell C recoveries (e.g., stronger immunity, less severe disease, spontaneous recovery). Critically, they rated these causes as less likely to be present on cell A trials, and those who attributed cell A recoveries only to the medicine showed the highest causal ratings for the medicine. Such a selective inference about the scope of a hidden cause may allow learners to explain away cue-absent outcomes that would otherwise undermine the perceived causal strength of the target cue, without violating normative principles. This mechanism may help explain how false causal beliefs can be maintained in the face of a null relationship.

  • Open Access

    Is generalization general? Differences between predictive and category learning reflect knowledge of “cognitive kinds”

    by René Schlegelmilch, Andy Wills, Jessica C. Lee

    Shepard (1987) observed that generalization gradients exhibited universal properties that were invariant between species and stimulus dimensions, and proposed that these invariances reflect knowledge of "natural kinds" that exist in the world. Here, we test a natural extension of Shepard's ideas, by examining whether generalization differs between different types of learned associations. To do so, we leverage generalization as a tool to diagnose the underlying content of learning in two domains (predictive learning and category learning). In two large experiments equating stimuli and procedural details between tasks, we show that empirical differences in gradients are largely explained by differences in the testing procedures (Experiment 1), and the degree to which outcomes and categories are encoded as mutually exclusive (Experiment 2); not the type of association (predictive vs. categorical) that is learned. Critically, our findings demonstrate that qualitative differences in gradient shape corresponding to rule- and similarity-based generalization are affected by testing procedure, but do not differ by learning domain. We conclude that predictive and category learning are subsumed by the same basic mechanisms, and that the differences in generalization observed between domains reflect knowledge about "cognitive kinds"; categories are created to be mutually exclusive while outcomes are not.