A double error dynamic asymptote model of associative learning.

Abstract

In this paper a formal model of associative learning is presented which incorporates representational and computational mechanisms that, as a coherent corpus, empower it to make accurate predictions of a wide variety of phenomena that so far have eluded a unified account in learning theory. In particular, the Double Error Dynamic Asymptote (DDA) model introduces: 1) a fully-connected network architecture in which stimuli are represented as temporally clustered elements that associate to each other, so that elements of one cluster engender activity on other clusters, which naturally implements neutral stimuli associations and mediated learning; 2) a predictor error term within the traditional error correction rule (the double error), which reduces the rate of learning for expected predictors; 3) a revaluation associability rate that operates on the assumption that the outcome predictiveness is tracked over time so that prolonged uncertainty is learned, reducing the levels of attention to initially surprising outcomes; and critically 4) a biologically plausible variable asymptote, which encapsulates the principle of Hebbian learning, leading to stronger associations for similar levels of cluster activity. The outputs of a set of simulations of the DDA model are presented along with empirical results from the literature. Finally, the predictive scope of the model is discussed.

Representing uncertainty in the Rescorla-Wagner model: Blocking, the redundancy effect, and outcome base rate

“…It would also be useful to know whether other associative models could account for these results (e.g. Gershman, 2015;Kokkola et al, 2019;McLaren & Mackintosh, 2000Schmajuk et al, 1996;Wagner, 1981); with or without modifications to the initial associative strengths.…”

The representation of stimulus conjunction in theories of associative learning: A context-dependent added-elements model.

“…In fact, across four experiments there were numerical differences in learning rate in the direction opposite to that predicted by McLaren and Mackintosh (2000, 2002). Recently, Kokkola, Mondragón, and Alonso (2019) described a novel elemental connectionist network model, which can account for a wide range of learning phenomena. It is possible that this model may also accommodate the results discussed here, but that so far remains to be confirmed.…”

Mediated learning: A computational rendering of ketamine-induced symptoms.

“…No data were used in this article. Additional details and formalisms of the DDA model were published in N. H. Kokkola et al (2019).…”

“…Associative learning has a longstanding influence on clinical studies and treatment and is still one of the main paradigms for understanding the basic principles of human pathological behavior (Corlett & Schoenbaum, 2021;Haselgrove & Hogarth, 2011;Lewis et al, 2013;Schachtman & Reilly, 2011). In this article, I present simulations of the DDA model (N. H. Kokkola et al, 2019) that aim to replicate some empirical manipulations that produce psychosislike effects in rodents. The DDA model, unlike other theories that encompass algorithms for associatively activated cue learning (e.g., Brandon et al, 2000;Wagner, 1981;Wagner & Brandon, 2001;but see McLaren, 2011;McLaren et al, 1989) incorporates computational mechanisms to integrate attention and learning for memory retrieved representations.…”