ABSTRACT
So far we have considered what is perhaps the simplest viable form of Γ within a system, where only U ⇒ a, but the remaining parameters were either (i) fixed, or (ii) varied randomly in time at a second-order level, or (iii) could be externally varied nonrandomly but independently of U. Case V 7 was not devised first (Gregson, 1984) but was considered here first because of its relatively direct relation to the CNO-generating property which is of interest in psychophysics. However, though V 7 is well-behaved in some ways, for that very reason it cannot be used to mimic some of the more ill-behaved phenomena listed in Al to A19 of Chapter 2. Also, the V 1 to be considered here is the starting point of a number of other variants which are readily produced by relaxing or removing constraints on the parameter bounds and on the maximal range of rates of change from one cycle to the next. The idea now is to set up the most complicated identifiable structure that might be plausibly studied, and then to consider it and a number of cases derived by simplifying or constraining the general case. This is not simply a mathematical device, it also suggests ways in which transient lesions in mapping between parts of a dynamic system can readily induce the sort of phenomena which are reported in experiments where there is a failure to obtain a monotonically increasing S — R psychometric function like a CNO.
