Descriptive or normative? Sample Clauses

Descriptive or normative?. The arguments in what follows will not always be clearly on one side of the norm- ative/descriptive divide. In general we will consider what a rational agent ‘ought’ to do (or might be best advised to do, which comes to the same thing); how- ever, given the avowed level of approximation required in the model, it could be somewhat unclear whether the outcomes of the theory are a recommendation for how ideal agents ought to behave, or how we might expect that they probably do behave. The theory we will develop could indeed be used in different ways along the de- scriptive/normative spectrum. At the descriptive end, we can see the theory as helping to explain why our evolution might have produced certain features and behaviours. We will see this most visibly cashed out in explanations of parts of linguistic behaviour and our mental categorisation process in Part II of the thesis. Towards the normative end, the image of rationality developed here is a depiction of optimality-under-constraint, which could (as suggested above) be seen as a yardstick for actual human behaviour. In a different vein, our findings could (in theory) be used to inform the design of an artificial intelligence, helping it to behave in an optimal manner within its computational constraints. To some extent, I am relatively relaxed about the possibilities here. Although the argument will use somewhat normative terminology, my own view of the theory’s predictions lies perhaps somewhere in the middle of the spectrum: it is more justificatory than normative as such. To the extent to which the theory will seem to explain an observable feature of human behaviour, we can potentially understand that behaviour in terms of the theory (descriptively),29 and can potentially view it as at least somewhat rational, with a degree of confidence determined by the level of approximation inherent in the original premises. The above being laid on the table, we will proceed in the next chapter to look at the basic model of memory, designed to fit within the general philosophical approach we wish to sell (that of memory as a rational design problem). The chapter will be as non-mathematical as possible, with the mathematical xxxx being held back until Chapter 4. 29This suggests the interesting question of how realist we can be about the model and its picture of memory as a decision process subject to rational control. Although we certainly do not engage in conscious cost–benefit calculations when encoding items,...