We are capable of visually picking out and keeping track of particular objects. You can visually differentiate a bird, and you can perceive it as remaining the same individual as it flies across your field of vision. Partly to explain this ability, many authors have posited the existence of an “object file” system, which is recruited when we pick out and track individual objects. Object files have also played an important role in some recent theories of perception-based singular thought. In his talk, Green will be concerned with the question of what object files pick out.

Several have proposed that the object file system selects and tracks in accordance with certain “object principles.” In particular, Tyler Burge and Susan Carey hold that the object file system is governed by the principles of*three-dimensionality*,*cohesion*, and*boundedness*. Green contrasts this approach with a view on which the visual system selects and tracks in accordance with familiar Gestalt criteria of perceptual organization. Green shows that the criteria of perceptual organization are, in general, far more permissive than the three-dimensionality, cohesion, and boundedness principles. Thus, Green calls the former position the*Restrictive View*, and the latter position the*Permissive View*.

Green argues that the arguments standardly adduced in support of the Restrictive View are unpersuasive, and that the evidence in fact better supports the Permissive View. Processes of selection and tracking appear not to impose a three-dimensionality requirement, and there is little evidence that they draw a sharp distinction between cohesive objects and perceptual groups. Green also argues that work on the perception of so-called “nonmaterial” entities, such as holes, supplies further support for the Permissive View over its competitor. Finally, Green considers a weaker version of the Restrictive View, on which the object file system has the function of selecting entities that satisfy three-dimensionality, cohesion, and boundedness, even though it does not critically rely on representations of these properties. Green will also raise doubts about this weaker position.