This paper provides an epistemological and methodological analysis of the recent practice of using neural language models to simulate brain language processing. It is argued that, on the one hand, this practice can be understood as an instance of the traditional simulative method in artificial intelligence, following a mechanistic understanding of the mind; on the other hand, that it modifies the simulative method significantly. Firstly, neural language models are introduced; a study case showing how neural language models are being applied (...) in cognitive neuroscience for simulative purposes is then presented; after recalling the main epistemological features of the simulative method in artificial intelligence, it is finally highlighted how the epistemic opacity of neural language models is tackled by using the brain itself to simulate the neural language model and to test hypotheses about it, in what is called here a co-simulation. (shrink)

This article sets in with the question whether current or foreseeable transformer-based large language models (LLMs), such as the ones powering OpenAI’s ChatGPT, could be language users in a way comparable to humans. It answers the question negatively, presenting the following argument. Apart from niche uses, to use language means to act. But LLMs are unable to act because they lack intentions. This, in turn, is because they are the wrong kind of being: agents with intentions need to be autonomous (...) organisms while LLMs are heteronomous mechanisms. To conclude, the article argues, based on structural aspects of transformer-based LLMs, that these LLMs have taken a first step away from mechanistic artificiality to autonomous self-constitution, which means that these models are (slowly) moving into a direction that someday might result in non-human, but equally non-artificial agents, thus subverting the time-honored Kantian distinction between organism and mechanism. (shrink)

After briefly laying out a cultural-evolutionary approach to speech acts (Sects. 1–2), I argue that the notion of commitment at play in assertion and related speech acts comprises multiple dimensions (Sect. 3). Distinguishing such dimensions enables us to hypothesize evolutionary precursors to the modern practice of assertion, and facilitates a new way of posing the question whether, and if so to what extent, speech acts are conventional (Sect. 4). Our perspective also equips us to consider how a modern speaker might (...) employ an illocutionary analogue of A.N. Prior’s “runabout-inference ticket”, in which the pragmatic “introduction rules” for utterances correspond to evolutionary precursors of modern speech acts, but in which the “elimination rules” correspond to their modern descendants (Sect. 5). Such behavior would be abusive, though not in a way readily discernible without an evolutionary perspective on speech acts that attends to the dimensions of commitment that they encompass. Such behavior also raises the question how we may safeguard against it in public discourse, and I close (Sect. 6) with some suggestions for doing so. (shrink)

We critically survey prominent recent scholarship on the question of whether fiction can be a source of epistemic value for those who engage with it fully and appropriately. Such epistemic value might take the form of knowledge (for ‘cognitivists’) or understanding (for ‘neo-cognitivists’). Both camps may be sorted according to a further distinction between views explaining fiction’s epistemic value either in terms of the author’s engaging in a form of telling, or instead via their showing some state of affairs to (...) obtain, a special case of which is the provision of self-knowledge. Fictional works that show rather than tell often employ thought experiments. The epistemic value of some fictional works is indicated by their enabling of empathy, itself illuminated via the psychological process of experience-taking. Whether a fictional work offers epistemic value by telling or showing, there is, in principle, no bar to its being able to deliver on what it offers, and consumers of fiction who exercise epistemic vigilance may gain either knowledge or some degree of understanding from their engagement with it. (shrink)

The world is witnessing a rise in speech-enabled devices serving as epistemic informants to their users. Some philosophers take the view that because the utterances produced by such machines can be phenomenologically similar to an equivalent human speech, and they may deliver the same function in terms of delivering content to their audience, such machine utterances should be conceptualized as “assertions”. This paper argues against this view and highlights the theoretical and pragmatic challenges faced by such a conceptualization which seems (...) to be based on a “functionalist” account of assertion. Instead, the paper argues that the view that some such machine utterances can be “proxy assertions”, made on behalf of the designers and/or deployers of such machines is more tenable. Existing accounts of such a view, however, need further refinement, which is provided here. The paper then discusses implications of this view for design of such machines, particularly those enabled and equipped with machine learning capabilities. (shrink)