Science, Logic, and Mathematics

Artificial neural networks and supervised learning have become an essential part of science. Beyond using them for accurate input-output mapping, there is growing attention to a new feature-oriented approach. Under the assumption that networks optimised for a task may have learned to represent and utilise important features of the target system for that task, scientists examine how those networks manipulate inputs and employ the features networks capture for scientific discovery. We analyse this approach, show its hidden caveats, and suggest its (...) legitimate use. We distinguish three things that scientists call a ‘feature’: parametric, diagnostic, and real-world features. The feature-oriented approach aims for real-world features by interpreting the former two, which also partially rely on the network. We argue that this approach faces a problem of non-uniqueness: there are numerous discordant parametric and diagnostic features and ways to interpret them. When the approach aims at novel discovery, scientists often need to choose between those options, but they lack the background knowledge to justify their choices. Consequentially, features thus identified are not promised to be real. We argue that they should not be used as evidence but only used instrumentally. We also suggest transparency in feature selection and the plurality of choices. (shrink)

In this research, patent prosecution is conceptualized as a system of reinforcement learning from human feedback. The objective of the system is to increase the likelihood for a language model to generate patent claims that have a higher chance of being granted. To showcase the controllability of the language model, the system learns from granted patents and pre-grant applications with different rewards. The status of “granted” and “pre-grant” are perceived as labeled human feedback implicitly. In addition, specific to patent drafting, (...) the experiments in this research demonstrate the model’s capability to learn from adjusting claim length and inclusion of limiting terms for narrowing claim scope. As proof of concept, the experiments focus on claim ones only and the training data originates from a patent dataset tailored specifically for artificial intelligence. Although the available human feedback in patent prosecution are limited and the quality of generated patent text requires improvement, the experiments following the 3-stage reinforcement learning from human feedback have demonstrated that generative language models are capable of reflecting the human feedback or intent in patent prosecution. To enhance the usability of language models, the implementation in this research utilizes modern techniques that enable execution on a single consumer-grade GPU. The demonstrated proof of concept, which reduces hardware requirements, will prove valuable in the future as more human feedback in patent prosecution become available for broader use, either within patent offices or in the public domain. (shrink)

This paper analyses whether current explainable AI (XAI) techniques can help to address taxpayer concerns about the use of AI in taxation. As tax authorities around the world increase their use of AI-based techniques, taxpayers are increasingly at a loss about whether and how the ensuing decisions follow the procedures required by law and respect their substantive rights. The use of XAI has been proposed as a response to this issue, but it is still an open question whether current XAI (...) techniques are enough to meet existing legal requirements. The paper approaches this question in the context of a case study: a prototype tax fraud detector trained on an anonymized dataset of real-world cases handled by the Buenos Aires (Argentina) tax authority. The decisions produced by this detector are explained through the use of various classification methods, and the outputs of these explanation models are evaluated on their explanatory power and on their compliance with the legal obligation that tax authorities provide the rationale behind their decision-making. We conclude the paper by suggesting technical and legal approaches for designing explanation mechanisms that meet the needs of legal explanation in the tax domain. (shrink)

In this essay I suggest that we view design principles in systems biology as minimal models, for a design principle usually exhibits universal behaviors that are common to a whole range of heterogeneous (living and nonliving) systems with different underlying mechanisms. A well-known design principle in systems biology, integral feedback control, is discussed, showing that it satisfies all the conditions for a model to be a minimal model. This approach has significant philosophical implications: it not only accounts for how design (...) principles explain, but also helps clarify one dispute over design principles, e.g., whether design principles provide mechanistic explanations or a distinct kind of explanations called design explanations. (shrink)

Information Retrieval systems benefit from the use of long queries containing a large volume of search-relevant information. This situation is not common, as users of such systems tend to use very short and precise queries with few keywords. In this work we propose a modification of the Latent Dirichlet Allocation (LDA) technique using data from the document collection and its vocabulary for a better representation of short queries. Additionally, a study is carried out on how the modification of the proposed (...) LDA weighted vectors increase the performance using relevant documents as feedback. The work shown in this paper is tested using three biomedical corpora (TREC Genomics 2004, TREC Genomics 2005 and OHSUMED) and one legal corpus (FIRE 2017). Results prove that the application of the proposed representation technique, as well as the feedback adjustment, clearly outperforms the baseline methods (BM25 and non-modified LDA). (shrink)

In this paper, we introduce the notions of connexive and bi-intuitionistic multilattices and develop on their base the algebraic semantics for Kamide, Shramko, and Wansing’s connexive and bi-intuitionistic multilattice logics which were previously known in the form of sequent calculi and Kripke semantics. We prove that these logics are sound and complete with respect to the presented algebraic structures.

Many philosophers have argued that statistical evidence regarding group characteristics (particularly stereotypical ones) can create normative conflicts between the requirements of epistemic rationality and our moral obligations to each other. In a recent article, Johnson-King and Babic argue that such conflicts can usually be avoided: what ordinary morality requires, they argue, epistemic rationality permits. In this article, we show that as data get large, Johnson-King and Babic’s approach becomes less plausible. More constructively, we build on their project and develop a (...) generalized model of reasoning about stereotypes under which one can indeed avoid normative conflicts, even in a big data world, when data contain some noise. In doing so, we also articulate a general approach to rational belief updating for noisy data. (shrink)

Recent trends in the debate about natural kinds tend towards increasingly permissive and practice-oriented views. I argue that while these accounts—which I characterize using Boyd’s ([1991]) term ‘enthusiasm’—offer several helpful insights, they often lack the normative force that they want to have; that is, they cannot provide an account of what makes something a good or bad, better or worse, kind for scientific pursuits. I argue that such accounts can regain a minimal sense of normativity in two ways. The first, (...) which has already been widely embraced by theorists, is to index the appropriateness of a scientific kind to the domain within which it is situated. While this is a good solution, it is not clear how it can be applied in interdisciplinary and cross-domain research pursuits, which are increasingly common. The second way of regaining normativity—which is especially, but not exclusively, effective for such cross-domain pursuits—is to attend to the stage at which scientific research is occurring. (shrink)

A core, constitutive norm of science is to remove or remedy the artefacts in one’s data. Here, we consider examples of artefacts from many fields of science (for example, astronomy, economics, electrophysiology, psychology, and systems neuroscience) and discuss their contribution to a more general evidential selection problem at the heart of the epistemology of evidence. Synthesizing and building on previously disparate discussions in many areas of the philosophy of science, we provide a novel, causal–pragmatic account that fits the examples and (...) clarifies the practices by which artefacts are discovered and remedied. (shrink)

Functionalism as a philosophical position has been recently applied to the case of emotion research. However, a number of objections have been raised against applying such a view to scientific theorizing on emotions. In this article, I argue that functionalism is still a viable strategy for emotion research. To do this, I present functionalism in philosophy of mind and offer a sketch of its application to emotions. I then discuss three recent objections raised against it and respond to each of (...) them. These objections claim that functionalism is intractable because (i) it does not support a scientifically interesting taxonomy of emotions for experimental settings, (ii) it is inherently teleological, and (iii) it cannot be falsified. I argue that these objections either rely on a simplified version of functionalism as a philosophical position or they pose challenges that functionalists can readily address. Lastly, I conclude by drawing some lessons these objections suggest for a tractable functionalist account of emotions. (shrink)

I argue that non-epistemic values are necessarily embedded in the measure of evidential strength of contrastive evidence. When evidence is contrastive, evidence is stronger the more it favours a hypothesis over a set of plausible, mutually exclusive alternative hypotheses. In such a contrastive epistemic setting, evidence has an effect not only on a particular hypothesis, but on the whole probability distribution over the set of alternative hypotheses. A natural way of analysing the incremental impact of new evidence on a set (...) of alternative hypotheses is in terms of uncertainty or ‘entropy’ reduction. There is no unique single measure of uncertainty/entropy, however, and, consequently, no single unique measure of uncertainty reduction. Finally, I argue that the ‘right’ measure of entropy reduction, and hence of evidential strength, depends on the pragmatic context and non-epistemic values. (shrink)

The reliability of previous findings on two crucial aspects of counterfactual thinking, namely the content of counterfactual modifications and their impact on future performance, has been questioned for the frequent use of tasks characterised by simple causal domains, that restrict participants’ possibility to consider a broad range of modifications. To overcome this limitation, we utilised a new experimental task featuring a complex causal domain to investigate such key aspects. The results indicated that participants tend to generate counterfactuals about elements outside (...) their control, especially when presented with a challenging version of the task (Study 1a) and, to a lesser extent, when they receive negative feedback on their performance (Study 1b). Moreover, despite occasional implementations of actions mentioned in controllable counterfactuals, being engaged in counterfactual thinking did not lead to subsequent performance improvements (Study 2). The implications of these findings for the debate on the function of counterfactual thinking are discussed. (shrink)