Concepts, Pointers and Occasion-Specific Thoughts

Concepts, Pointers and Occasion-Specific Thoughts
3 June 2019, 10.30am - 4 June 2019, 4.30pm
Room G7, Ground Floor, Senate House, Malet Street, London WC1E 7HU

This conference is about concepts, word meanings, and the rich bodies of information that guide human thought. 


Emma Borg (Reading) "Do word meanings fix denotations and truth conditions or not?" 

Chris Eliasmith (Waterloo) "Concepts as discrete and continuous representations and processes"
Charles Gallistel (Rutgers) "It's numbers all the way down" 
Ekaterini Klepousniotou (Leeds) "Processing and representation of polysemy”  
Jake Quilty Dunn (Oxford/WUSTL) "Concepts as generative pointers" 
Nicholas Shea (Institute of Philosophy) "Three ways of linking information to a concept"

Long-term memory includes sentence-like representations (e.g. 1000mm equals one metre), direct inferential links between concepts (e.g. cat -> mammal), and sets of sensory, motor, motivational and affective representations (e.g. a visual image of a rolled-up piece of lawn turf). We can call this large collection the ‘conception’ connected with a concept. Not all conceptions are deployed on each occasion when we reason or think with a particular concept, but each is important on certain occasions. How do concepts relate to conceptions? 

We want to explore the idea that conceptions are organized around a ‘label’ that can be used in thought. Thinkers can think with the label without using the conception it is connected with (Fodor 1998). Perhaps labels can be concatenated into structured representations, such that JANE PETS FIDO is different from FIDO PETS JANE in virtue of label-like representations arranged differently. We could then perform logical inferences that are sensitive only to these logical forms. 

How do these conceptual labels link up with conceptions? An idea explored here is that they function computationally as “pointers.” Pointers are representations that redirect access elsewhere in memory. Perhaps conceptual labels are pointers to conceptions.  

How should we think about pointers? Gallistel & King (2009) set out the classical computational idea of a pointer. It is an address that names a memory location where relevant data is stored. Computations can be performed on the pointer without calling the stored data. Chris Eliasmith (2013) has argued for the existence of ‘semantic pointers’, which operate in a different way. A semantic pointer is a compressed representation which can be decompressed into detailed sensorimotor information. Here, too, operations can be performed on a pointer without calling the more specific representations to mind.  

Concepts are often thought to function as (representations of) word meanings. Word meanings seem to be polysemous—that is, they flexibly generate senses from a single core meaning, as ‘chicken’ can refer to an animal or the meat from that animal. Some evidence suggests that polysemy is distinct from homonymy, in which a word like ‘bear’ has multiple unrelated meanings (Klepousniotou et al., 2012; MacGregor et al. 2015). In that case, perhaps a single conceptual label/pointer can generate different denotations and truth conditions on different occasions. This family of issues raises some key questions that will be discussed at this conference. 

Question 1: Is the relation between labels and conceptions usefully understood in terms of a pointer architecture? 

Question 2: Can structure-sensitive inferences (and other operations) be performed over labels/pointers independently of conceptions? 

Question 3: How do we use conceptual pointers to construct occasion-specific thoughts? 

Question 4: What are word meanings, such that they can generate these occasion-specific thoughts (Carston 2012)? And do they fix denotations and truth conditions (Borg 2004) or not (Pietroski 2018)? 

Question 5: Do polysemous words map to a single concept (Vicente 2018), multiple concepts (Murphy 2002), or are word meanings nonconceptual (Pietroski 2018)? 

The support of the Institute of Philosophy and the European Research Council (MetCogCon project GA681422) is gratefully acknowledged.


Emily Gardner
020 7862 8867