Elemental representation and configural mappings: Combining elemental and configural theories of associative learning

Abstract

In this paper we present our first attempt at combining an elemental theory designed to model representation development in an associative system (e.g., McLaren, Kaye and Mackintosh, 1989), with a configural theory that models associative learning and memory (McLaren, 1993). After considering the possible advantages of such a combination (and some possible pitfalls), we offer a hybrid model that allows both components to produce the phenomena that they are capable of without introducing unwanted interactions. We then successfully apply the model to a range of phenomena including latent inhibition, perceptual learning, the Espinet effect, and first and second order retrospective revaluation. In some cases we present new data for comparison with our model results; in all cases the model replicates the pattern observed in our experimental results. We conclude that this line of development is a promising one for generating truly comprehensive theories of learning and memory. IntroductionOur long-term aim is to produce a general model of associative learning and memory that captures the processes that are common to both humans and infra-humans. This paper investigates the feasibility of combining elemental and configural approaches to associative learning and memory as a stepping stone towards that ultimate goal. In doing so, it also addresses a particular computational problem: of how we can have representation development at an elemental level whilst still learning in a holistic fashion. We start by considering the problem in general, and by motivating the need to find a solution incorporating both elemental (e.g., Estes, 1959;Mclaren, Kaye and Mackintosh, 1989;Brandon, Vogel and Wagner, 2000;Harris, 2006) and configural (e.g., Pearce, 1987, McLaren, 1993Honey, 2000; Honey and WardRobinson, 2002) forms of representation, then move to a specific example of such a combination that attempts to amalgamate the McLaren, Kaye and Mackintosh (1989, henceforth MKM) model of representation development with the APECS (McLaren, 1993;Le Pelley and McLaren, 2001;McLaren, 2011) model of associative learning and memory. To anticipate slightly, the enterprise is a successful one in the sense that the hybrid model is able to reproduce the phenomena that can be simulated using its components (and thus is of wider scope than either of its constituent parts), but this outcome was not achieved without considerable effort and overcoming numerous difficulties. In the course of grappling with this problem we have developed a new respect for the way in which issues multiply as the complexity of the model increases, and we try to pass on our experience of what will and will not work when synthesizing elemental and configural approaches to learning and memory.

The effect of disrupting configural information on rats’ performance in the Morris water maze

“…They argued that these results were consistent with the proposition that a change in the stimulus conditions from the training phase to the test phase led to generalization decrement. There is no doubt that Pearce's (1987) theory is one of those capable of providing both effects via one similarity-based mechanism (for other theories capable of generating this result, see McLaren, Forrest, & McLaren, 2012;and Honey, 2000). We also agree that a simple elemental theory employing something like the Rescorla-Wagner algorithm (Rescorla & Wagner, 1972) would have to appeal to a process such as external inhibition in combination with overshadowing to explain this result, making it a less plausible account of these results.…”

Associations and propositions: The case for a dual-process account of learning in humans

“…Thus, where participants were able to use the propositional system to solve the problem, they did not show an irrational unblocking effect (as we would expect), but where they had to rely on associative processes, they did. Another set of experiments that support this thesis using a similar manipulation (though this time with second order retrospective revaluation as the effect in question) can be found in McLaren, Forrest and McLaren (2012).…”

Is performance in task-cuing experiments mediated by task set selection or associative compound retrieval?

“…In order to demonstrate this we modeled the data using the most recent version of the APECS model, a three-layer backpropagation localist connectionsist network -see Figure 8, McLaren, (2011) andMcLaren, Forrest &McLaren (2012) for more details of the model. The main differences between APECS and a standard back-propagation network are that APECS has adaptive learning parameters which favor the unit best suited for carrying a given mapping, and adaptive biases to preserve this learning, and that it is trained to replicate the input on the output units as well as generate the right response.…”