Reasons For Bias Research Articles

Bias Sensitivity in Diagnostic Decision-Making: Comparing ChatGPT with Residents.

Diagnostic errors, often due to biases in clinical reasoning, significantly affect patient care. While artificial intelligence chatbots like ChatGPT could help mitigate such biases, their potential susceptibility to biases is unknown. This study evaluated diagnostic accuracy of ChatGPT against the performance of 265 medical residents in five previously published experiments aimed at inducing bias. The residents worked in several major teaching hospitals in the Netherlands. The biases studied were case-intrinsic (presence of salient distracting findings in the patient history, effects of disruptive patient behaviors) and situational (prior availability of a look-alike patient). ChatGPT's accuracy in identifying the most-likely diagnosis was measured. Diagnostic accuracy of residents and ChatGPT was equivalent. For clinical cases involving case-intrinsic bias, both ChatGPT and the residents exhibited a decline in diagnostic accuracy. Residents' accuracy decreased on average 12%, while the accuracy of ChatGPT 4.0 decreased 21%. Accuracy of ChatGPT 3.5 decreased 9%. These findings suggest that, like human diagnosticians, ChatGPT is sensitive to bias when the biasing information is part of the patient history. When the biasing information was extrinsic to the case in the form of the prior availability of a look-alike case, residents' accuracy decreased by 15%. By contrast, ChatGPT's performance was not affected by the biasing information. Chi-square goodness-of-fit tests corroborated these outcomes. It seems that, while ChatGPT is not sensitive to bias when biasing information is situational, it is sensitive to bias when the biasing information is part of the patient's disease history. Its utility in diagnostic support has potential, but caution is advised. Future research should enhance AI's bias detection and mitigation to make it truly useful for diagnostic support.

Multi-stage reasoning on introspecting and revising bias for visual question answering

Visual Question Answering (VQA) is a task that involves predicting an answer to a question depending on the content of an image. However, recent VQA methods have relied more on language priors between the question and answer rather than the image content. To address this issue, many debiasing methods have been proposed to reduce language bias in model reasoning. However, the bias can be divided into two categories: good bias and bad bias. Good bias can benefit to the answer prediction, while the bad bias may associate the models with the unrelated information. Therefore, instead of excluding good and bad bias indiscriminately in existing debiasing methods, we proposed a bias discrimination module to distinguish them. Additionally, bad bias may reduce the model’s reliance on image content during answer reasoning and thus attend little on image features updating. To tackle this, we leverage Markov theory to construct a Markov field with image regions and question words as nodes. This helps with feature updating for both image regions and question words, thereby facilitating more accurate and comprehensive reasoning about both the image content and question. To verify the effectiveness of our network, we evaluate our network on VQA v2 and VQA cp v2 datasets and conduct extensive quantity and quality studies to verify the effectiveness of our proposed network. Experimental resu- lts show that our network achieves significant performance against the previous state-of-the-art methods.

Problem Behaviours and Relinquishment: Challenges Faced by Clinical Animal Behaviourists When Assessing Fear and Frustration.

Fear and frustration are two emotions thought to frequently contribute to problem behaviour, often leading to relinquishment. Inferring these emotions is challenging as they may present with some similar general signs, but they potentially require different treatment approaches to efficiently address the behaviour of concern. Although behavioural assessment frameworks have been proposed, it is largely unknown how clinical animal behaviourists (CABs) assimilate information about the emotional state of an animal to inform their behavioural assessment. In other fields (such as both in human and veterinary medicine), the use of intuition and gut feelings, without the concurrent use of an assessment framework, can lead to higher rates of error and misdiagnosis. Therefore, this study used semi-structured interviews of ten CABs and qualitative methods to explore the ways they conceptualise, recognise and differentiate fear and frustration in dogs. Although interviewees perceived fear and frustration as negative affective states that lead to changes in an animal's behaviour, there was little consensus on the definition or identification or differentiation of these emotions. The use of a scientific approach (i.e., hypothesis-driven and based on falsification of competing hypotheses) for behavioural assessment was highly variable, with individual assessment processes often characterised by tautology, intuition, circular reasoning and confirmation bias. Assessment was typically based on professional judgment, amalgamating information on interpretation of communicative signals, motivation, learning history, breed, genetics and temperament. Given the lack of consensus in the definition of these states, it is clearly important that authors and clinicians define their interpretation of key concepts, such as fear and frustration, when trying to communicate with others.

Post Hoc Bias in Treatment Decisions

A goal of health care is to reduce symptoms and improve health status, whereas continuing dubious treatments can contribute to complacency, discourage the search for alternatives, and lead to shortfalls in care. To test a potential bias in intuitive reasoning following a marginal improvement in symptoms after a dubious treatment (post hoc bias). Surveys eliciting treatment recommendations for hypothetical patients were sent to community members throughout North America recruited via an online survey platform in the early winter months of 2023 and 2024 and presented to health care professionals (pharmacists who were approached in person using a secret shopper technique) in the summer months of 2023. Respondents received randomized versions of surveys that differed according to whether vague symptoms improved or remained unchanged after a dubious treatment. The primary outcome was a recommendation to continue treatment. In total, 1497 community members (mean [SD] age, 38.1 [12.5] years; 663 female [55.3%]) and 100 health care professionals were contacted. The first scenario described a patient with a sore throat who took unprescribed antibiotics; respondents were more likely to continue antibiotics after initial treatment if there was a marginal improvement in symptoms vs when symptoms remained unchanged (67 of 150 respondents [45%] vs 25 of respondents [17%]; odds ratio [OR], 3.98 [95% CI, 2.33-6.78]; P < .001). Another scenario described a patient with wrist pain who wore a copper bracelet; respondents were more likely to continue wearing the copper bracelet after initial care was followed by a marginal improvement in symptoms vs when symptoms remained unchanged (78 of 100 respondents [78%] vs 25 of 99 respondents [25%]; OR, 16.19 [95% CI, 5.32-19.52]; P < .001). A third scenario described a patient with fatigue who took unprescribed vitamin B12; respondents were more likely to continue taking vitamin B12 when initial treatment was followed by a marginal improvement in symptoms vs when symptoms remained unchanged (80 of 100 respondents [80%] vs 33 of 100 respondents [33%]; OR, 7.91 [95% CI, 4.18-14.97]; P < .001). Four further scenarios involving dubious treatments found similar results, including when tested on health care professionals. In this study of clinical scenarios, a marginal improvement in symptoms led patients to continue a dubious and sometimes costly treatment, suggesting that clinicians should caution patients against post hoc bias.

Easy-fix attentional focus manipulation boosts the intuitive and deliberate use of base-rate information.

In addressing human reasoning biases, "easy-fix" attentional focus interventions have shown that we can prompt reasoners to align responses with logico-mathematical principles. The current study aimed to test the impact of such interventions on both intuitive and deliberate responses on base-rate items. Using a two-response paradigm, participants provided initial intuitive responses under time constraints and cognitive load, followed by deliberate responses. During the intervention, we used attentional focus manipulations with base-rate items that aimed to redirect participants' attention toward the "logical" base-rate cue (i.e., the logical intervention) or toward the "heuristic" descriptive cue (i.e., the heuristic intervention). The results indicate that the logical intervention led to improved alignment with logico-mathematical principles in both intuitive and deliberate responses, albeit with a modest effect size. Conversely, the heuristic intervention had no discernible impact on accuracy. This indicates that our attentional focus manipulation is more effective at getting reasoners to respect rather than to override base-rates.

Clinical decision making within occupational therapy: a critical reflection of providing palliative care for a patient in the community

Clinical decision-making within occupational therapy is a complex and dynamic process shaped by numerous scientific, narrative, professional, ethical, legal and personal factors. By reflecting on these factors and by developing metacognition of one's own reasoning pathways, biases, and heuristics, patient-centred and balanced clinical decisions can be made. This reflective case study examines the factors underpinning an occupational therapist's clinical decision making for a man with terminal cancer. A critical reflective model is used in this article to provide structure and to ensure critical consideration of the case. Implications for practice aim to serve as inspiration for others to reflect on their own journey, and personal action points are detailed to develop the author's own clinical decision-making.

Bayes versus bias in human reasoning.

Bayes versus bias in human reasoning.

The roots of polarization in the individual reward system.

Polarization raises concerns for democracy and society, which have expanded in the internet era where (mis)information has become ubiquitous, its transmission faster than ever, and the freedom and means of opinion expressions are expanding. The origin of polarization however remains unclear, with multiple social and emotional factors and individual reasoning biases likely to explain its current forms. In the present work, we adopt a principled approach and show that polarization tendencies can take root in biased reward processing of new information in favour of choice confirmatory evidence. Through agent-based simulations, we show that confirmation bias in individual learning is an independent mechanism and could be sufficient for creating polarization at group level independently of any additional assumptions about the opinions themselves, a priori beliefs about them, information transmission mechanisms or the structure of social relationship between individuals. This generative process can interact with polarization mechanisms described elsewhere, but constitutes an entrenched biological tendency that helps explain the extraordinary resilience of polarization against mitigating efforts such as dramatic informational change in the environment.

Role of knowledge and reasoning processes as predictors of resident physicians’ susceptibility to anchoring bias in diagnostic reasoning: a randomised controlled experiment

BackgroundDiagnostic errors have been attributed to reasoning flaws caused by cognitive biases. While experiments have shown bias to cause errors, physicians of similar expertise differed in susceptibility to bias. Resisting...

The relevance of the contribution of psychoneuroendocrinoimmunology and psychology of reasoning and decision making to nursing science: A discursive paper.

Patients' death or adverse events appear to be associated with poor healthcare decision-making. This might be due to an inability to have an adequate representation of the problem or of the connections among problem-related elements. Changing how a problem is formulated can reduce biases in clinical reasoning. The purpose of this article is to explore the possible contributions of psychoneuroendocrinoimmunology (PNEI) and psychology of reasoning and decision-making (PRDM) to support a new nursing theoretical frame. Discursive paper. This article discusses the main assumptions about nursing and nurses' ability to face patient's problems, suggesting a new approach that integrates knowledge from PNEI and PRDM. While PNEI explains the complexity of systems, highlighting the importance of systems connections in affecting health, PRDM underlines the importance of the informative context in creating a mental representation of the problem. Furthermore, PRDM suggests the need to pay attention to information that is not immediately explicit and its connections. Nursing recognizes the patient-nurse relationship as the axiom that governs care. The integration of PNEI and PRDM in nursing theoretics allows the expansion of the axiom by providing essential elements to read a new type of relationship: the relationship among information. PNEI explains the relationships between biological systems and the psyche and between the whole individual and the environment; PRDM provides tools for the nurse's analytical thinking system to correctly process information and its connections. A theoretical renewal is mandatory to improve nursing reasoning and nursing priority identification. Integrating PNEI and PRDM into nursing theoretics will modify the way professionals approach patients, reducing cognitive biases and medical errors. There was no patient or public involvement in the design or writing of this discursive article.

Teleological reasoning bias is predicted by pupil dynamics: Evidence for the extensive integration account of bias in reasoning.

Teleological reasoning is the tendency for humans to see purpose and intentionality in natural phenomena when there is none. In this study, we assess three competing theories on how bias in reasoning arises by examining performance on a teleological reasoning task while measuring pupil size and response times. We replicate that humans (N = 45) are prone to accept false teleological explanations. Further, we show that errors on the teleological reasoning task are associated with slower response times, smaller baseline pupil size, and larger pupil dilations. The results are in line with the single-process extensive integration account and directly oppose predictions from dual-processing accounts. Lastly, by modeling responses with a drift-diffusion model, we find that larger baseline pupil size is associated with lower decision threshold and higher drift rate, whereas larger pupil dilations are associated with higher decision threshold and lower drift rate. The results highlight the role of neural gain and the Locus Coeruleus-Norepinephrine system in modulating evidence integration and bias in reasoning. Thus, teleological reasoning and susceptibility to bias likely arise due to extensive processing rather than through fast and effortless processing.

The desirability bias in personality-related syllogistic reasoning.

The belief-bias effect is a tendency to evaluate syllogistic statements based on believability rather than on formal logic validity. Following this rationale, the study examines desirability bias as the tendency to evaluate syllogistic conclusions based on their desirability when reasoning is conducted on personality-relevant categorical syllogisms. For this purpose, 60 syllogisms were constructed based on the items of the Big Five questionnaire. Syllogisms were subsequently categorized as desirable (e.g., "I empathize with others") and undesirable (e.g., "I am passive") based on their conclusion. In each task, the second premise and the conclusion were formulated in the first person to increase a respondent's identification with the content. A total of 300 university students (Mage = 20.08, SD = 2.02) participated in the study. A 2 (syllogism validity: valid, invalid) × 2 (syllogism desirability: desirable, undesirable) repeated measures ANOVA was employed. The analysis showed a greater tendency to accept desirable conclusions on valid syllogisms (valid desirable rather than valid undesirable) and reject undesirable conclusions on invalid syllogisms (invalid undesirable rather than invalid desirable). These findings have implications for socially desirable responding in cognitive tasks, which may be further developed as a source of self-relevant content as well as for further extension of belief bias in the form of desirability bias.

Iconicity bias and duration.

Descriptions of durational relations can be ambiguous, for example, the description "one meeting happened during another" could mean that one meeting started before the other ended, or it could mean that the meetings started and ended simultaneously. A recent theory posits that people mentally simulate descriptions of durational events by representing their starts and ends along a spatial axis, that is, an iconic representation of time. To draw conclusions from this iconic mental model, reasoners consciously scan it in the direction of earlier to later timepoints. The account predicts an iconicity bias: People should prefer descriptions that are congruent with an iconic scanning procedure-descriptions that mention the starts of events before the ends of events-over logically equivalent but incongruent descriptions. Six experiments corroborated the prediction; they show that iconicity biases in temporal reasoning manifest in cases when reasoners consciously evaluate the durations of events. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Revisiting the narrow latent scope bias in explanatory reasoning

Humans are capable explainers and lay people tend to share the same explanatory virtues held in high regard by philosophers and scientists. However, a recent line of studies found a striking deviation from normativity in lay people’s explanations, termed the “narrow latent scope bias”. When competing explanations with identical a priori probabilities fit observed evidence equally well – but differ in the number of unobserved pieces of evidence they predict (latent scope) – reasoners seem to prefer explanations that predict fewer unobserved pieces of evidence (narrow latent scope). This tendency has been described as a robust explanatory reasoning bias. The present paper empirically demonstrates across six experiments (N=2200) that this bias is less robust than has been claimed, and influenced by nuanced pragmatic inferences on the side of participants. Pragmatic factors shown to influence the bias are assumptions about how easily an unobserved piece of evidence should have been observed if it was present (“feature diagnosability”), and the formulation of the test question being asked. Across studies, genuine narrow latent scope biases resulting from fallacious reasoning were found only in a fraction of participants. It is also demonstrated that the magnitude of the bias depends on response options: it is stronger if participants are forced to commit an error, but at best weak if they are allowed to give the correct answer.

Judicial Production of Racial Injustice in Taiwo v Olaigbe: Decolonising the Incomplete Story on Race and Contracting

In Taiwo, one of the most recent landmark cases on racial justice, the Supreme Court rejected race discrimination claims of two domestic migrant workers, ruling that discrimination on the basis of ‘immigration status’ should not be equated to discrimination on the basis of ‘race’. This article presents an argument for decolonising judicial decision-making, using Taiwo as an example to reimagine a much more favourable outcome for victims of racial injustice. This argument is explored through three propositions for decolonial judgment: (a) challenging racial bias in judicial reasoning and legal doctrine; (b) challenging legal frameworks as sites of racial oppression and inequality; and (c) accounting for contextual diversity of experiences of racialisation, avoiding essentialist arguments and categories of racial discrimination. Drawing on these, the article retells the stories in Taiwo to challenge the dominant, traditional race equality paradigm and expose the varied and multi-layered ways in which people are racialised differently across historical and socio-cultural contexts and communities. It also opens the potential for an epistemic shift away from the liberal paradigm of ‘freedom of contract’ and towards the analysis of racial contracting that is co-constituted by multi-layered and context-situated structures of oppression and domination.

Psychosocial processes and human desire: an inconvenient truth about online misinformation

Much has been written about the potential civic ramifications of online misinformation, and scholars have identified many useful strategies for helping students discern fact from fiction on social media. However, those strategies make an assumption, which is that consumers of digital media have a desire to identify and share accurate information. In this article, I argue that media literacy efforts should also require students to be retrospective and grapple with their own motivations and biases that may make them more susceptible to believing inaccurate or misleading information. Drawing on theories from political psychology, namely motivated reasoning and confirmation bias, as well as psychosocial research that explains the role of affect on individuals’ decision-making, I identify implications for K-12 media literacy education.

Mitigating cognitive bias with clinical decision support systems: an experimental study

ABSTRACT Bias in clinical reasoning has been identified as one of the main sources of diagnostic errors. Clinical Decision Support Systems that suggest possible diagnoses and provide information to mitigate cognitive bias could support physicians in finding a less biased diagnosis. We examine the influence of confidence and experience on the probability to adjust the decision after receiving decision aid and whether forming a first opinion beforehand or immediately receiving decision support makes a difference. 103 physicians and medical students participated in an online experiment built on decision tasks formulated to trigger availability and representativeness bias. The analysis showed that the presentation of prevalence data to mitigate availability bias changed the final probability estimate of the diagnosis significantly. Prototypical data to counteract representativeness bias showed no significant change. Medical experience, confidence in the decision, and timing of support had no significant influence on the probability to change the estimate.

The policy acknowledgement gap: Explaining (mis)perceptions of government social program use

AbstractNearly, all Americans have received social policy benefits, yet many do not acknowledge “using government social programs.” Why? Work on the submerged state proposes that people who receive social assistance through market mechanisms do not realize that the benefits they get are the result of government policy, and therefore, they do not acknowledge receiving government assistance. Others point to motivated reasoning or social desirability bias to explain the gap between acknowledging and using social programs. We classify the existing literature into three broad explanations—delivery, definition, and desirability—and propose that each may be responsible for people's inability to accurately report using government social programs. We test these mechanisms with original survey experiments. The results of this study provide support for the theory that multiple mechanisms are at work in shaping social policy acknowledgment, but they confirm that a partisan acknowledgement gap exists across a variety of conditions, and it persists despite treatments designed to minimize it. The study has significant implications for the conditions under which partisanship and policy usage coalesce to undermine support for government social expenditures, and it helps to explain the persistence of a “makers vs. takers” logic in American politics.

Cognitive biases in first-episode psychosis with and without attention-deficit/hyperactivity disorder.

Psychotic disorders such schizophrenia and attention-deficit/hyperactivity disorder (ADHD) are neurodevelopmental disorders with social cognitive deficits. Specifically, biased interpretation of social information can result in interpersonal difficulties. Cognitive biases are prevalent in psychosis, but no previous study has investigated whether the type and severity of cognitive biases differ between subjects experiencing first-episode psychosis (FEP) with (FEP-ADHD+) and without ADHD (FEP-ADHD-). A total of 121 FEP outpatients at the Early Intervention Service of Reus were screened for childhood ADHD through the Diagnostic Interview for ADHD (DIVA). Cognitive biases were assessed by the Cognitive Biases Questionnaire for Psychosis (CBQp). CBQp scores of FEPs groups were compared with those of healthy controls (HCs) with an analysis of covariance. Spearman correlation analysis explored associations between CBQp scores and psychopathology. Thirty-one FEPs met the criteria for childhood ADHD and reported significantly more cognitive bias [median (interquartile range): 47 (38-56)] than FEP-ADHD- [42 (37-48)] and HCs [38 (35.5-43)]. CBQp scores did not differ between FEP-ADHD-and HCs when adjusted for age and sex. After controlling for clinical differences, Intentionalising (F = 20.97; p < 0.001) and Emotional Reasoning biases (F = 4.17; p = 0.04) were more strongly associated with FEP-ADHD+ than FEP-ADHD-. Cognitive biases were significantly correlated with positive psychotic symptoms in both groups but only with depressive symptoms in FEP-ADHD- (r = 0.258; p = 0.03) and with poor functioning in FEP-ADHD+ (r = -0.504; p = 0.003). Cognitive bias severity increased from HCs to FEP-ADHD-patients to FEP-ADHD+ patients. FEP-ADHD+ patients may be a particularly vulnerable group in which metacognitive targeted interventions are needed.

An analysis of how motivated reasoning and confirmation bias affect people to accept “true news” through social media selectively

In recent years, with the development of technology, social media has started to appear in people’s lives, rapidly entering people’s lives along with the progress of the Internet. However, with the rise of social media, self-publishing has gradually proliferated on the Internet, with all kinds of news being sent without verification, causing much distress to people. It is, therefore, essential to understand why people selectively accept “true news” through social media. This study examines why people believe in fake news from the perspective of different “Emotional” Tendencies through a content analysis of four posts from different social media applications. The findings show that the public relies on theories related to motivated reasoning and confirmation bias when selectively accepting fake news. In particular, confirmation bias is a case where people, keen to believe in claims that are consistent with their beliefs or their lack of knowledge, are often attracted to the words and texts of the poster and unconsciously believe their story, thus contributing to their agenda. This research studies some of the main factors that have contributed to the public’s selective belief in fake news in recent years and helps to understand more about how people perceive and recognize fake news.