Logical Fallacies and Misinterpretations that Hinder Progress in Translational Addiction Neuroscience

Substance use disorders (SUDs) are heterogenous and complex, making the development of translationally predictive rodent and non-human primate models to uncover their neurobehavioral underpinnings difficult. Neuroscience-focused outcomes have become highly prevalent, and with this, the notion that SUDs are disorders of the brain embraced as a dominant theoretical orientation to understand SUD etiology and treatment. These efforts, however, have led to few efficacious pharmacotherapies, and in some cases (as with cocaine or methamphetamine), no pharmacotherapies have translated from preclinical models for clinical use. In this review and theoretical commentary, we first describe the development of animal models of SUDs from a historical perspective. We then define and discuss three logical fallacies including 1) circular explanation, 2) affirming the consequent, and 3) reification that can apply to developed models. We then provide three case examples in which conceptual or logical issues exist in common methods (i.e., behavioral economic demand, escalation, and reinstatement). Alternative strategies to refocus behavioral models are suggested for the field in an attempt to better bridge the translational divide between animal models and the clinical condition of SUDs.

On logical inference over brains, behaviour, and artificial neural networks

In the cognitive, computational, and neuro- sciences, we often reason about what models (viz., formal and/or computational) represent, learn, or "know", as well as what algorithm they instantiate. The putative goal of such reasoning is to generalize claims about the model in question to claims about the mind and brain. This reasoning process typically presents as inference about the representations, processes, or algorithms the human mind and brain instantiate. Such inference is often based on a model's performance on a task, and whether that performance approximates human behaviour or brain activity. The model in question is often an artificial neural network (ANN) model, though the problems we discuss are generalizable to all reasoning over models. Arguments typically take the form "the brain does what the ANN does because the ANN reproduced the pattern seen in brain activity" or "cognition works this way because the ANN learned to approximate task performance." Then, the argument concludes that models achieve this outcome by doing what people do or having the capacities people have. At first blush, this might appear as a form of modus ponens, a valid deductive logical inference rule. However, as we explain in this article, this is not the case, and thus, this form of argument eventually results in affirming the consequent – a logical or inferential fallacy. We discuss what this means broadly for research in cognitive science, neuroscience, and psychology; what it means for models when they lose the ability to mediate between theory and data in a meaningful way; and what this means for the logic, the metatheoretical calculus, our fields deploy in high-level scientific inference.

The Knowledge Problem in Mature Science

In the course of the nineteenth century, physical scientists became increasingly self-conscious of the need for a theory of how scientific knowledge was produced. Though many theories were proposed, none won a consensus. As explicitly stated by William Whewell, the core problem was the same for everyone: how to ground claims of knowledge of experience in a way that also justified claiming that the object of these claims was a reality independent of experience that caused experience. Everyone was acutely aware of the Fallacy of Affirming the Consequent and of the logical gulf between induction and deduction. John Herschel, Whewell, John Stuart Mill, August Comte, Hermann Helmholtz, Pierre Duhem, and Ernst Mach were some who proposed theories of science. Of these, Mach alone decisively rejected reality as the objective of science. Meanwhile, the nonscientist J. B. Stallo argued for the fundamental role played by metaphysical concepts in modern science.

Sensitivity of implicit evaluations to accurate and erroneous propositional inferences

Explicit (directly measured) evaluations are widely assumed to be sensitive to logical structure. However, whether implicit (indirectly measured) evaluations are uniquely sensitive to co-occurrence information or can also reflect logical structure has been a matter of theoretical debate. To test these competing ideas, participants (N = 3,928) completed a learning phase consisting of a series of two-step trials. In step 1, one or more conditional statements (A → B) containing novel targets co-occurring with valenced adjectives (e.g., “if you see a blue square, Ibbonif is sincere”) were presented. In step 2, a disambiguating stimulus, e.g., blue square (A) or gray blob (¬A) was revealed. Co-occurrence information, disambiguating stimuli, or both were varied between conditions to enable investigating the unique and joint effects of each. Across studies, the combination of conditional statements and disambiguating stimuli licensed different normatively accurate inferences. In Study 1, participants were prompted to use modus ponens (inferring B from A → B and A). In Studies 2–4, the information did not license accurate inferences, but some participants made inferential errors: affirming the consequent (inferring A from A → B and B; Study 2) or denying the antecedent (inferring ¬B from A → B and ¬A; Studies 3A, 3B, and 4). Bayesian modeling using ordinal constraints on condition means yielded consistent evidence for the sensitivity of both explicit (self-report) and implicit (IAT and AMP) evaluations to the (correctly or erroneously) inferred truth value of propositions. Together, these data suggest that implicit evaluations, similar to their explicit counterparts, can reflect logical structure.

Sensitivity of implicit evaluations to accurate and erroneous propositional inferences

Explicit (directly measured) evaluations are widely assumed to be sensitive to logical structure. However, whether implicit (indirectly measured) evaluations are uniquely sensitive to co-occurrence information or can also reflect logical structure has been a matter of theoretical debate. To test these competing ideas, participants (N = 3,928) completed a learning phase consisting of a series of two-step trials. In step 1, one or more conditional statements (A → B) containing novel targets co-occurring with valenced adjectives (e.g., “if you see a blue square, Ibbonif is sincere”) were presented. In step 2, a disambiguating stimulus, e.g., blue square (A) or gray blob (¬A) was revealed. Co-occurrence information, disambiguating stimuli, or both were varied between conditions to enable investigating the unique and joint effects of each. Across studies, the combination of conditional statements and disambiguating stimuli licensed different normatively accurate inferences. In Study 1, participants were prompted to use modus ponens (inferring B from A → B and A). In Studies 2–4, the information did not license accurate inferences, but some participants made inferential errors: affirming the consequent (inferring A from A → B and B; Study 2) or denying the antecedent (inferring ¬B from A → B and ¬A; Studies 3A, 3B, and 4). Bayesian modeling using ordinal constraints on condition means yielded consistent evidence for the sensitivity of both explicit (self-report) and implicit (IAT and AMP) evaluations to the (correctly or erroneously) inferred truth value of propositions. Together, these data suggest that implicit evaluations, similar to their explicit counterparts, can reflect logical structure.

Logical Intuition Is Not Really About Logic

Recent research suggest that reasoners are able to draw simple logical or probabilistic inferences relatively intuitively and automatically, a capacity which has been termed “logical intuition” (see, for example, De Neys & Pennycook, 2019). A key finding in support of this interpretation is that conclusion validity consistently interferes with judgments of conclusion believability, suggesting that information about logical validity is available quickly enough to interfere with belief judgments. In this paper we examined whether logical intuitions arise because reasoners are sensitive to the logical features of problem or another structural feature that just happens to aligns with logical validity. In three experiments (N = 113, 137, and 122), we presented participants with logical (determinate) and pseudo-logical (indeterminate) arguments and asked them to judge the validity or believability of the conclusion. Logical arguments had determinately valid or invalid conclusions, whereas pseudo-logical arguments were all logically indeterminate, but some were pseudo-valid (possible ‘strong’ arguments) and others pseudo-invalid (possible ‘weak’ arguments). Experiments 1 and 2 used simple Modus Ponens and Affirming the Consequent structures; Experiment 3 used more complex Denying the Antecedent and Modus Tollens structures. In all three experiments, we found that pseudo-validity interfered with belief judgments to the same extent as real validity. Altogether, these findings suggest that whilst people are able to draw inferences intuitively, and these inferences impact on belief judgments, they are not ‘logical intuitions.’ Rather, the intuitive inferences are driven by the processing of more superficial structural features that happen to align with logical validity.

Evolution in the Valley of Illusions

This letter discusses the relationship between fitness and fitness landscapes in evolution, showing the difficulty in finding optimal solutions. The idea that modern biology must be the result of evolution because it is fit is shown to be a case of affirming the consequent.

Logical Fallacies as Bayesian Informational Shortcuts

Information closure may help with the consistency of a database, so it is related to information quality. However, it cannot be used to expand such an information repository. For this, other forms of reasoning are needed. Bayesianism is often indicated as a classic means to upgrade a set of beliefs or indeed some bits of information, in the vocabulary of this book. Some other erroneous forms of reasoning, however, damage the same reservoir of information. Interestingly, the two dynamics are related. As argued in this chapter, the two best known formal logical fallacies, namely denying the antecedent (DA) and affirming the consequent (AC), are not just basic and simple errors, which prove human irrationality, but rather informational shortcuts, which may provide a quick and dirty (and therefore unsafe) way of extracting useful information from the same informational resources to which Alice already has access. And, in this sense, they can be shown to amount to degraded versions of Bayes’ theorem, once this is stripped of some of its probabilities. The less the probabilities count, the closer these fallacies become to a reasoning that is not only informationally useful but also logically valid.

The Urgent Need of a Naturalized Logic

The naturalization of logic aims at a revision of mainstream logic. In this article, I contend it is an urgent task to be completed. This new project will permit a new collaboration between logic and cognitive science. This can be accomplished doing for logic what many decades ago Quine and other philosophers undertook in the case of epistemology. First of all, this article analyzes how the naturalization can be achieved thanks to some insights provided by the recent John Woods’ book Errors of Reasoning: Naturalizing the Logic of Inference; important concepts that regard a naturalized logic are synthetically analyzed: errors (and the problem of fallacies), paradigm creep, third-way reasoning, consequence-having and consequence drawing, agent based reasoning. The article also takes advantage of my own studies, which are aimed both at exculpating the negative fallacious character of abduction (it is the fallacy of the affirming the consequent) and at illustrating the EC-model (Eco-Cognitive model) of it, I have recently proposed. Aiming at encouraging the project of naturalization of logic, the article specifically recommends the increase of logical research on abduction, and emphasizes how current philosophical and logical research on human inferences is indebted towards Charles Sanders Peirce, a philosopher whose importance and modernity are too often underestimated. The final part of the article will introduce an analysis of the importance of the so-called optimization of situatedness, a concept that is necessary to understand that maximization of “abducibility”, which characterizes modern science.