Gabriel Segal

The topic of this paper is the semantic structure of belief reports of the form 'a believes that p'. it is argued that no existing theory of these sentences satisfactorily accounts for anaphoric relations linking expressions within the embedded complement sentence to expressions outside. a new account of belief reports is proposed which assigns to embedded expressions their normal semantic values but which also exploits frege's idea of using senses to explain the apparent failures of extensionality in the reports

Allow me to recapitulate some territory that will be familiar to most readers. Here is how the problem of mental causation has typically been set up since shortly after the onset of non-reductive physicalism. It is now widely assumed that the realm of the physical is causally closed: every physical event has a complete physical cause, a cause that is sufficient for the event’s occurrence. This apparently leaves us with a limited number of options concerning psychological causation, none of which appear hugely attractive. Either: (a) the psychological is epiphenomenal and can have no causal impact on the physical, or (b) the psychological is identical with the physical, or (c) thoughts and actions are all over-determined, each one having two distinct sufficient causes. Option (b) subdivides into two further options. Either (b1) the psychological reduces to the physical and every psychological property is identical with some physical property, or (b2) token psychological events are identical with or constituted from token physical events but psychological properties are not identical with physical properties. (b1) is widely held to be inconsistent with the multiple realisation of the psychological by the physical. And (b2) appears to bring us back to the original problematic, with the properties as the locus of tension. If one event causes another it does so in virtue some of its properties and not others. If I throw a stone at a window and the window breaks, it is because the stone was hard and heavy that it broke the window and not, say, because it was grey and millions of years old. The properties in virtue of which an event has a particular effect are typically called the ‘causally efficacious properties of the cause with respect to the effect.’ Suppose, then that token neural event causes an action. We can ask ‘Does it do so in virtue of its physical properties or its psychological properties?’ and we are back to choosing between options (a) and (c) or returning to (b1)..

Please allow me to recapitulate some territory that will be familiar to most readers. Here is how the problem of mental causation has typically been set up since shortly after the onset of non-reductive physicalism. It is now widely assumed that the realm of the physical is causally closed. This means that the probability of any event’s occurring is fully determined by physical causes, and physical causes alone. There is no space in the physical causal nexus for any non-physical event to exert any influence

The paper begins with the assumption that psychological event tokens are identical to or constituted from physical events. It then articulates a familiar apparent problem concerning the causal role of psychological properties. If they do not reduce to physical properties, then either they must be epiphenomenal or any effects they cause must also be caused by physical properties, and hence be overdetermined. It then argues that both epiphenomenalism and over‐determinationism are prima facie perfectly reasonable and relatively unproblematic views. The paper proceeds to argue against Kim’s (Kim, 2000, 2005) attempt to articulate a plausible version of reductionism. It is then argued that psychological properties, along with paradigmatically causally efficacious macro‐properties, such as toughness, are causally inefficacious in respect of their possessor’s typical effects, because they are insufficiently distinct from those effects. It is finally suggested that the distinction between epiphenomenalism and overdeterminationism may be more terminological than real.