Epistemic Error Research Articles

Cross-fade sampling: extremely efficient Bayesian inversion for a variety of geophysical problems

SUMMARY This paper introduces cross-fade sampling, a computationally efficient Markov Chain Monte Carlo simulation method that uses a semi-analytical approach to quickly solve Bayesian inverse problems that do not themselves have an analytical solution. Cross-fading is efficient in two ways. First, it requires fewer samples to obtain the same quality simulation of the target probability density function (PDF). Secondly, it is much faster to evaluate the posterior probability of each sample than conventional sampling methods for simulating Bayesian posterior PDFs. Conventional methods require evaluating the prior probability (which describes your a priori constraints) and data likelihood (which describes the fit between the observations and the predictions of the model) for each sample model. However, cross-fading does not require evaluating the data likelihood, meaning that ‘big data’ can be fit with zero additional computational cost. Further, the cross-fading approach can be used to calculate the marginal likelihood associated with a model design, facilitating model comparison and Bayesian model averaging. Topics covered in this paper include derivation of the cross-fade approach and how it can be used to simulate Bayesian posterior PDFs and compute the marginal likelihood, discussion of the class of problems to which cross-fading can be applied (with examples from earthquake statistics, earthquake ground motion modelling, volcanic eruption forecasting, and finite fault slip modelling), demonstration of efficiency relative to existing sampling methods and discussion of how cross-fading can be used to account for prediction errors (i.e. epistemic errors) as part of the geophysical inverse problem.

Evaluation of hydrological models at gauged and ungauged basins using machine learning-based limits-of-acceptability and hydrological signatures

Hydrological models are evaluated by comparisons with observed hydrological quantities such as streamflow. A model evaluation procedure should account for dominantly epistemic errors in hydrological data such as model input precipitation and streamflow and avoid type-2 errors (rejecting a good model). This study uses quantile random forest (QRF) to develop limits-of-acceptability (LoA) over streamflows that account for uncertainties in precipitation and streamflow values. A significant advantage of this method is that it can be used to evaluate models even at ungauged basins. This method was used to evaluate a hydrological model –Sacramento Soil Moisture Accounting (SAC-SMA) – over the St. Joseph River Watershed (SJRW) for both gauged and hypothetical ungauged scenarios. QRF defined wide LoAs that yielded a large number of models as behavioral, suggesting the need for additional measures to develop a more discriminating inference procedure. The paper discusses why the LoAs defined by QRF were wide, along with some ways to define more discriminating LoAs. To further constrain the model, five streamflow-based signatures (i.e., autocorrelation function, Hurst exponent, baseflow index, flow duration curve, and long-term runoff coefficient) were used. The combination of LoAs over streamflow and streamflow-based signatures helped constrain the set of behavioral models in both the gauged and the ungauged scenarios. Among the signatures used in this study, the Hurst exponent and baseflow index were the most useful ones. All the 1-million models evaluated in this study were eventually rejected as unfit-for-purpose.

UPH Problem 20 – reducing uncertainty in model prediction: a model invalidation approach based on a Turing-like test

Abstract. This study proposes using a Turing-like test for model evaluations and invalidations based on evidence of epistemic uncertainties in event runoff coefficients. Applying the consequent “limits of acceptability” results in all the 100 000 model parameter sets being rejected. However, applying the limits, together with an allowance for timing errors, to time steps ranked by discharge, results in an ensemble of 2064 models that can be retained for predicting discharge peaks. These do not include any of the models with the highest (> 0.9) efficiencies. The analysis raises questions about the impact of epistemic errors on model simulations, and the need for both better observed data and better models.

The time validity of Philip's two‐term infiltration equation: An elusive theoretical quantity?

AbstractThe two‐term infiltration equation is commonly used to determine the sorptivity, , and product, , of the dimensionless multiple and saturated soil hydraulic conductivity from cumulative vertical infiltration measurements (L) at times (T). This reduced form of the quasi‐analytical power series solution of Richardson's equation of Philip enjoys a solid physical underpinning but at the expense of a limited time validity. Using simulated infiltration data, Jaiswal et al. have shown this time validity to equal about 2.5 cm of cumulative infiltration. The goals of this work are twofold. First, we investigate the extent to which cumulative infiltration measurements larger than 2.5 cm bias the estimates of and . Second, we investigate the impact of epistemic errors on the inferred time validities and parameters. Partial infiltration curves up to 2.5 cm of cumulative vertical infiltration improve substantially the agreement between actual and least squares estimates of and . But this only holds if the data generating infiltration process follows Richardson's equation and experimental conditions satisfy assumptions of soil homogeneity and a uniform initial water content. Otherwise, autocorrelated cumulative infiltration residuals will bias the least squares estimates of and . Our findings reiterate and reinvigorate earlier conclusions of Haverkamp et al. and show that epistemic errors deteriorate the physical significance of the coefficients of infiltration functions. As a result, the parameters of infiltration functions cannot simply be used in storm water and vadose zone flow models to forecast runoff and recharge at field and landscape scales unless these predictions are accompanied by realistic uncertainty bounds. We conclude that the time validity of Philip's two‐term equation is an elusive theoretical quantity with arbitrary physical meaning.

Sensor placement for data assimilation of turbulence models using eigenspace perturbations

We present an approach to sensor placement for turbulent mean flow data assimilation in the context of Reynolds-averaged Navier–Stokes (RANS) simulations. It entails generating a spatial uncertainty map through the eigenspace perturbations (ESPs) of the baseline turbulence model (e.g., the k−ω shear stress transport model) to quantify the epistemic structural errors in the model. A novel greedy search algorithm is proposed to place sensors targeting regions of highest uncertainty in the spatial uncertainty map generated from ESP. The algorithm is computationally efficient (e.g., computational cost negligible compared to a RANS solution) and is both easy to implement and tune. It involves two hyper-parameters (a constraint to avoid sensor clustering and the number of sensors) which we investigate in-depth. A variational (adjoint-based) data assimilation approach is used for flow reconstruction. The proposed strategy was tested on three two-dimensional wall-bounded flows (Reynolds number ranging 5.6×103–9.36×105) involving flow separation and reattachment. For the wall-mounted hump case, we found that data assimilation using 33 sensors with our proposed sensor placement algorithm reduced the average velocity prediction error by 60% vs 38% with a simple uniform placement of sensors. Furthermore, we found that we could achieve 61% error reduction using our algorithm with only three sensors. Notably, in all tested cases, the error reduction using our method for sensor placement was close in accuracy to the instances where the entire flow field data were used for flow reconstruction, which involved two to three orders of magnitude more data points than the placed sensors.

Effects of aleatoric and epistemic errors in reference data on the learnability and quality of NN-based potential energy surfaces

The effect of noise in the input data for learning potential energy surfaces (PESs) based on neural networks for chemical applications is assessed. Noise in energies and forces can result from aleatoric and epistemic errors in the quantum chemical reference calculations. Statistical (aleatoric) noise arises for example due to the need to set convergence thresholds in the self consistent field (SCF) iterations whereas systematic (epistemic) noise is due to, inter alia, particular choices of basis sets in the calculations. The two molecules considered here as proxies are H2CO and HONO which are examples for single- and multi-reference problems, respectively, for geometries around the minimum energy structure. For H2CO it is found that adding noise to energies and forces with magnitudes representative of single-point calculations does not deteriorate the quality of the final PESs whereas increasing the noise level commensurate with electronic structure calculations for more complicated, e.g. metal-containing, systems is expected to have a more notable effect. On the other hand, for HONO which requires a multi-reference treatment, a clear correlation between model quality and the degree of multi-reference character as measured by the T1 amplitude is found. It is concluded that for chemically “simple” cases the effect of aleatoric and epistemic errors is manageable without evident deterioration of the trained model, but more care needs to be exercised for situations in which multi-reference effects are present.

The Importance of Realism about Gender Kinds: Lessons from Beauvoir

Abstract Beauvoir’s The Second Sex stands out as a master class in the accommodation of conceptual and inferential practices to real, objective gender kinds. Or so I will argue. To establish this framing, we will first need in hand the kind of scientific epistemology that correctly reconciles epistemic progress and error, particularly as pertains to the unruly social sciences. An important goal of the paper is to develop that epistemological framework and unlock its ontological implications for the domain of gender. As we will see, the real gender kinds that contemporary social scientists successfully identify and track are very much the same kinds to which Beauvoir was coordinating reference in The Second Sex. The correct identification of those kinds endures as a moral and political priority, regardless of one’s other gender-related normative agenda.

Leibniz's Moral Psychology of an Evil Person

Abstract Our focus in this article concerns Leibniz's views on evil. Our goal is to examine which are the consequences of his conception of moral agency for the moral psychology of the genuinely evil person. For Leibniz, moral failure is an epistemic error since it involves some false practical judgement. Moral maxims may be represented in blind or symbolic cognitions, but then moral agents can misrepresent the evil consequences of their behaviour. Finally, we discuss Leibniz's view on habits that may help virtuous persons strive for perfection but also enable evil persons to continue sinning.

A Probative Argument of Intention

This article develops the basic argument for proving intention in criminal adjudication. The evidential proof of an accused's state of mind is a very old problem that arises from an even older moral quandry of distinguishing between unlawful killings that are accidental from those that are intentional for punitive purposes. Intentional unlawful killings typically attract much greater responsibility and punishment than killings by accident. The necessity of proving intention for an unlawful killing arises generally from an alleged killer claiming: ‘I didn’t mean to do it’. This denial can in turn be disambiguated into two cognitive (epistemic error) and volitional (muscular accident) sub-categories. Historically, pre-modern fact-finders saw themselves as being limited to relying on confessions and expert marabout to prove an alleged killer's intention in the face of these denials. The article deploys contemporary rationalist methods of logical argumentation for the proof of intention in criminal adjudication. The arguments made in the paper are pitched at a theoretical level and without necessarily reflecting a detailed exposition of the existing doctrine of one particular jurisdiction.

Epistemic agency and the generalisation of fear

Fear generalisation is a psychological phenomenon that occurs when fear that is elicited in response to a frightening stimulus spreads to similar or related stimuli. The practical harms of pathological fear generalisation related to trauma are well-documented, but little or no attention has been given so far to its epistemic harms. This paper fills this gap in the literature. It shows how the psychological phenomenon, when it becomes pathological, substantially curbs the epistemic agency of those who experience the fear that generalizes, limiting their ability to respond to evidence, and substantially limiting their epistemic horizons. It is argued that when these epistemic harms are caused by wrongful actions and decisions of individuals or institutions, because the fear is elicited in response to a traumatic experience inflicted by them, the harms should be considered epistemic wrongs. The epistemic wrongs are closely akin to agential epistemic injustice, a variety of distributive epistemic injustice, and sometimes also involve epistemic exclusion. The paper thereby identifies a previously underexplored psychological mechanism that can be a vehicle through which both individuals and institutions can epistemically wrong others. The argument has implications for how both epistemic wrongs and epistemic injustice should be conceived, suggesting that both can occur without being caused by primarily epistemic flaws or errors, or a bad epistemic character. Finally, it highlights the advantage of taking a victim-centred approach to understanding epistemic harm.

The Perils of Rejecting the Parity Argument

AbstractMany moral error theorists reject moral realism on the grounds that moral realism implies the existence of categorical normativity, yet categorical normativity does not exist. Call this the Metaphysical Argument. In response, some moral realists have emphasized a parity between moral normativity and epistemic normativity. They argue that if one kind of normativity is rejected, then both must be rejected. Therefore, one cannot be a moral error theorist without also being an epistemic error theorist. Call this the Parity Argument. In this paper, we address three possible responses to the Parity Argument from moral error theorists: (1) accept the parity but still reject epistemic error theory, (2) reject the parity, (3) accept the parity and defend epistemic error theory. We argue that there are problems with each of these responses, so the Parity Argument stands as a strong counterargument to the Metaphysical Argument. We conclude by drawing some lessons for any future challenges to the Parity Argument.

Hercules: Deep Hierarchical Attentive Multilevel Fusion Model With Uncertainty Quantification for Medical Image Classification

The automatic and accurate analysis of medical images (e.g., segmentation,detection, classification) are prerequisites for modern disease diagnosis and prognosis. Computer-aided diagnosis (CAD) systems empower accurate and effective detection of various diseases and timely treatment decisions. The past decade witnessed a spur in deep learning (DL)-based CADs showing outstanding performance across many health care applications. Medical imaging is hindered by multiple sources of uncertainty ranging fromnteasurement (aleatoric) errors, physiological variability, and limited medical knowledge (epistemic errors). However, uncertainty quantification (UQ) in most existing DL methods is insufficiently investigated, particularly in medical image analysis. Therefore, to address this gap, in this article, we propose a simple yet novel hierarchical attentive multilevel feature fusion model with an uncertainty-aware module for medical image classification coined <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Hercules</i> . This approach is tested on several real medical image classification challenges. The proposed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Hercules</i> model consists of two main feature fusion blocks, where the former concentrates on attention-based fusion with uncertainty quantification module and the latter uses the raw features. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Hercules</i> was evaluated across three medical imaging datasets, i.e., retinal OCT, lung CT, and chest X-ray. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Hercules</i> produced the best classification accuracy in retinal OCT (94.21%), lung CT (99.59%), and chest X-ray (96.50%) datasets, respectively, against other state-of-the-art medical image classification methods.

Effect of uncertainties on block volume estimation

The combination of the aleatory nature of the rock mass structure and the epistemic errors related to the survey methods make rock mass characterization a challenge despite the remarkable evolution of the survey tools and the research on the subject. In particular, significant uncertainties affect block volume estimation: the need for simplification connected to the engineering approach to rockfall problems, for instance, risks to mask the ripple effect of uncertainties on the reliability of the results. Even considering a simplified shape of the block created by three sets of discontinuities (i.e., a prism), the uncertainties on the geometrical characteristics of the discontinuities (orientation, spacing, and persistence) greatly influence the resulting volume distribution. It is a fact that a single value of the volume cannot be representative of the rock mass: the In Situ Block Size Distribution (IBSD) should be built to describe the variability of block volumes. Many statistical distribution functions can be used for fitting spacing data (i.e., gamma, negative exponential, log-normal, Weibull). The choice of the function must follow a rigorous evaluation of the goodness of fit. This research aims to assess the influence of the uncertainties related to the discontinuities sets, with particular reference to spacing samples, on block volume estimation. Through numerical examples and a case study, this research shows that a reduction of uncertainty can be reached by rigorous statistical processing of the data.

Relocating English Studies and SoTL in the Global South: Towards Decolonizing English and Critiquing the Coloniality of Language

South Africa has policies and frameworks for curriculum design, transformation, and quality assurance in each public institution of higher education (HE). These policies influence the scholarship of teaching and learning (SoTL), particularly at the departmental and disciplinary levels of English Studies. Despite the policy narratives and rhetoric, English Studies still carries vestiges of colonialism and apartheid in South Africa. Similarly, in other disciplines, scholars in the Global South have highlighted coloniality, epistemicides, epistemic errors, and epistemic injustices, but not in a dual critique of SoTL and the English language. Hypercritical self-reflexivity by academics should be the norm in SoTL, and this should be linked to language-based curriculum reforms and module content designs. All of these self-reflexive efforts should foreground how the mission to transform and decolonize is entangled with Eurocentric paradigms of English language teaching. This paper characterizes the nexus between SoTL and the coloniality of language within South African higher education. It also discusses and critiques the nature of an English department in a post-apartheid and postcolonial South Africa. In addition, it critiques the coloniality of language and imperial English language paradigms often embraced by higher education institutions (HEIs) in South Africa, and delineates curriculum transformation, Africanization, and decolonizing English within this educational sector. Finally, the paper challenges Eurocentric SoTL practices and colonialist English language paradigms by framing its argument within a critical southern decolonial perspective and a post-Eurocentric SoTL.

On (in)validating environmental models. 1. Principles for formulating a Turing‐like Test for determining when a model is fit‐for purpose

AbstractModel invalidation is a good thing. It means that we are forced to reconsider either model structures or the available data more closely, that is to challenge our fundamental understanding of the problem at hand. It is not easy, however, to decide when a model should be invalidated, when we expect that the sources of uncertainty in environmental modelling will often be epistemic rather than simply aleatory in nature. In particular, epistemic errors in model inputs may well exert a very strong control over how accurate we might expect model predictions to be when compared against evaluation data that might also be subject to epistemic uncertainties. We suggest that both modellers and referees should treat model validation as a form of Turing‐like Test, whilst being more explicit about how the uncertainties in observed data and their impacts are assessed. Eight principles in formulating such tests are presented. Being explicit about the decisions made in framing an analysis is one important way to facilitate communication with users of model outputs, especially when it is intended to use a model simulator as a ‘model of everywhere’ or ‘digital twin’ of a catchment system. An example application of the concepts is provided in Part 2.

Deception in the era of digital technologies and the distortion of reality and facts: An X-Ray of Nigerian peculiarities

The phenomenal increase in the level of digital technologies has in no small measure enabled and opened new vista of possibilities that were hitherto inconceivable before the advent of these digital technologies. Indeed, this phenomenon has also increased the ‘crescendo’ of various seemingly impossible trends in the area of circulation of information in the globalised world in ways such as global rapid communication, unhindered and constant access to information, democratised production and dissemination of information and digital content, and the ability to coordinate global political activities or movements through several populist strategies and political antics employed by politicians. Currently, it is somewhat seemed intractable to separate truth from falsehood due to striking similarities in the appearance of the two. However, these phenomena have brought about untoward development resulting in the amplification of various forms of digital deceptions peddled on several digital devices. Like other countries, Nigeria, too, is battling with the rise in populism politics, fake news, ethnic nationalism (laden with ethnic jingoism), hate and dangerous speech and other kinds of digital deceptions, among others. The escalating herder-farmer communal clashes, ethno-religious crises in some states in Nigeria, a tsunami of misinformation and disinformation about COVID-19, as well as political tensions between the two dominant political parties (All Progressives Congress and People’s Democratic Party (PDP)), have been influenced by deceptions, untruths and political propaganda. Therefore, through extensive interaction with (and review of) both the extant and current literature, this paper: provides further insights into the evolving issues regarding deceptions, fake news, epistemic errors as well as democratic and social harms emanating from deceptions in the Nigerian polity. The methodology used in this paper involves a comprehensive but selective literature review conducted to locate papers (Journal articles and theses or dissertations) on the foregoing evolving issues. Also, literature related to each of the analyses of the trend of fake news and other forms of digital deceptions in the light of the emerging post-truth era and their potential impacts on the country is incorporated into the review. However, a conclusion is drawn from the findings while the paper recommends that curtailing the influence of digital deceptions and fake news on the body polity of Nigeria requires collective responsibility of all the stakeholders coupled with enhanced steps (which include but are not limited to: credibility, media skills, regulation, collaboration, media literacy, professionalism, gatekeeping, self-censorship as well as detection) towards fighting the menace and morbidity of digital deceptions in the country.

Uncertainties: An investigation of aleatory and epistemic errors in market segmentation analysis

Despite the popularity of cluster analysis as a segmentation tool, its limitations continue to include the production of random solutions and the existence of uncertainties. This study aims to assist marketers in understanding the characteristics of festival goers based on music events in Malaysia. The present study investigates the existence and effect of uncertainties produced in cluster analysis results by using an artificial neural network (ANN). Four market segments are identified: the alarm hitter, the technology ticker, the plug puller, and the fuse blower. Error analysis results reveal that uncertainties may cause incorrect predictions. Academically, the limitations in existing market segmentation studies are highlighted by adding the process of ANN training and testing the segments generated from the cluster analysis. From the industry perspective, this approach introduces an important segmentation basis—technographic segmentation—to tap into the wired generation. Future research may extend this study and apply a nonprobabilistic neural network to eliminate the existence of errors in cluster analysis.

Correction to: Epistemic errors in students’ four fundamental operations of whole numbers: a radical constructivist grounded theory (RCGT) approach

Correction to: Epistemic errors in students’ four fundamental operations of whole numbers: a radical constructivist grounded theory (RCGT) approach

On Natural and Transcendental Illusions in a Kantian-Pragmatist Philosophical Anthropology

Abstract The covid-19 pandemic and the increasingly polarized political situation in many countries today have highlighted the significance of various humanly natural intellectual mistakes, cognitive biases, and widespread inferential errors. This essay examines, at a philosophical meta-level, the relation between our natural epistemic errors and the kind of humanly unavoidable transcendental illusion analyzed by Immanuel Kant in the Transcendental Dialectic of the First Critique. While both kinds of illusion are usually primarily discussed in an epistemological context, my approach is not exclusively epistemic. Rather, my main argument extends the broadly Kantian critique of reason – as a critical via negativa focusing on what is epistemically “wrong” about us, as the kind of cognitive agents we human beings naturally are – into a pragmatically enriched investigation of “the negative” as constitutive of the human condition, that is, our lack of sufficiently critical epistemic and ethical capacities, including our propensity to both cognitive error and moral evil. Our being inescapably inclined to manifest such negativities in our epistemic and ethical lives, and our insufficient awareness of doing so, are, arguably, “transcendental facts” about us, to be critically examined in terms of a Kantian-cum-pragmatist philosophical anthropology. This significantly enriches our picture of transcendental philosophy.

When do non-epistemic values play an epistemically illegitimate role in science? How to solve one half of the new demarcation problem

Solving the “new demarcation problem” requires a distinction between epistemically legitimate and illegitimate roles for non-epistemic values in science. This paper addresses one ‘half’ (i.e. a sub-problem) of the new demarcation problem articulated by the Gretchenfrage: What makes the role of a non-epistemic value in science epistemically illegitimate? I will argue for the Explaining Epistemic Errors (EEE) account, according to which the epistemically illegitimate role of a non-epistemic value is defined via an explanatory claim: the fact that an epistemic agent is motivated by a non-epistemic value explains why the epistemic agent commits a particular epistemic error. The EEE account is inspired by Douglas’ and Steel’s “functionalist” or “epistemic constraint” accounts of epistemic illegitimacy. I will suggest that the EEE account is able to meet two challenges that these two accounts face, while preserving the key intuition underlying both accounts. If my arguments succeed, then the EEE account provides a solution to one half of the new demarcation problem (by providing a definition of what makes the role of a non-epistemic value epistemically illegitimate) and it opens up new ways for addressing the other half (i.e. characterizing an epistemically legitimate role for non-epistemic values).