Видео

Abstract: The purpose of this talk is to argue for the possibility that embodied expressions of confidence reflect deeply plausible candidates for bridging the gap between a highly prominent phylogenetic origin story about capacities for self-knowledge (i.e. evaluative metacognition, cf. Proust 2013) and a prominent ontogenetic origin story about the same (i.e. mindreading from the social-scaffolding viewpoint, cf. Vygotsky 1934/1978; Heyes et al. 2020). The chief concern in the debate over metacognition is whether the evaluative species of metacognition lays the foundation for the metarepresentational species, commonly referred to as 'mindreading'. Specifically, it is about whether the cognitive systems of non-human animals that perform well in the so-called ‘uncertainty monitoring paradigm’ (e.g. Smith et al. 1995; Shields et al. 1997; Smith et al. 2003, 2006, 2008, 2012, 2013, 2014, 2018, 2019) and the like (e.g. Roberts et al. 2012; Iwasaki et al. 2013) possess significant phylogenetic cognitive ingredients to distinctly human capacities for acquiring self-knowledge. The deeper philosophical question is whether theories that appeal to mechanisms for monitoring, evaluating, predicting, and controlling cognitive processes might ever satisfactorily explain the origin of capacities for acquiring beliefs about the self qua beliefs about the self, i.e. theories based on the Bayesian Brain Hypothesis (Knill and Pouget 2004; Chater et al. 2010; Meyniel et al. 2015a; Hu et al. 2021) and the predicative processing framework (e.g. Friston and Stefan 2009; Clark 2013, 2015, 2016; Hohwy 2013, 2020). In this talk, I claim that Bayesian theories of metacognition (e.g. Meyniel et al. 2015a, 2015b; Hu et al. 2021) have the tools to explain the origin of distinctly human capacities for self-knowledge, but such an explanation demands an appeal to embodied cognition (cf. Clark 1999; Wilson 2002; Seth 2013; Wong 2018; Shapiro 2019), specifically embodied expressions of confidence (and uncertainty), which are particularly well-suited to explain the phylogenetic origin of self-knowledge, so long as the social-scaffolding view is on the right track to explain its ontogenetic origin.

I'm reminded about how John Stuart Mill responded to objections to determinism in human decision making. If determinism (necessitarianism) meant that there was some special extra thing that necessitated human actions over and above all the usual antecedents including what people wanted to do, then he'd deny that too, but determinism is just the view that all the antecedents taken together have a certain consequent. As he said it is not the view that there is some special further cause that draws along the effect like a magnet. Likewise consciousness is not something separate and over and above all the mental goings on but just the sum total of that. A way I think about the success problem is if the robot says it is conscious and I don't believe the robot why do I think the robot is lying? I use the word lying because it is more provocative, but I mean to include things like being unreliable. If I know the robot is honest and reliable (not prone to ChatGPT style hallucinations for example) and I know all its mechanisms and no there is no mechanism by which it could lie or be unreliable then how could I justify not believing its pronouncements. It seems like I can't imagine an honest reliable robot saying its conscious without it being conscious...

Thank you for posting this presentation. Lots to think about and digest.

"promo sm"