HeiDI: A model for Pavlovian learning and performance with reciprocal associations

Abstract

<jats:title>Abstract</jats:title><jats:p>Associative treatments of how Pavlovian conditioning affects conditioned behavior are rudimentary: A simple ordinal mapping is held to exist between the strength of an association (V) between a conditioned stimulus (CS) and an unconditioned stimulus (US; i.e., V<jats:sub>CS-US</jats:sub>) and conditioned behavior in a given experimental preparation. The inadequacy of this simplification is highlighted by recent studies that have taken multiple measures of conditioned behavior: Different measures of conditioned behavior provide the basis for drawing opposite conclusions about V<jats:sub>CS-US</jats:sub>. Here, we develop a simple model involving reciprocal associations between the CS and US (V<jats:sub>CS-US</jats:sub> and V<jats:sub>US-CS</jats:sub>) that simulates these qualitative individual differences in conditioned behavior. The new model, HeiDI (How excitation and inhibition Determine Ideo-motion), enables a broad range of phenomena to be accommodated, which are either beyond the scope of extant models or require them to appeal to additional (learning) processes. It also provides an impetus for new lines of inquiry and generates novel predictions.</jats:p>

Individual differences in the nature of conditioned behavior across a conditioned stimulus: Adaptation and application of a model.

“…then the values returned for RUS tend to be larger than those returned for RCS (cf. Honey et al, 2020a). One interesting (supplementary) observation is that in each case while the values returned for RCS increase according to a negatively accelerated function, those returned for RUS increase according to a sigmoidal function.…”

Elaboration of a model of Pavlovian learning and performance: HeiDI.

“…This analysis shows that -even when A-B associations do not play a role -HeiDI is able to explain results taken to be inconsistent with the use of the pooled error term by the Rescorla-Wagner model (see also, Holmes, Chan, & Westbrook, 2019). The analysis is straightforward, described in greater detail in Honey et al (2019), and can be verified using an open source app containing the code for the HeiDI model: https://ynnna.shinyapps.io/HeiDI_model/.…”

“…(is weaker) and A-US2 (is negative) after downshift unblocking. This analysis is formally presented in Honey et al (2019), and can also be verified using the open source HeiDI app.…”

Individual variation in the vigor and form of Pavlovian conditioned responses: Analysis of a model system

“…Learning ceases when ΣV TOTAL-CS = c.α CS , and the learning rate parameter β US affects the rate at which V US-CS approaches c.α CS . If we now consider what happens on a trial in which a compound of two stimuli (A and B) precedes a US, then the c.α CS values for each CS in a compound (i.e., c.α A and c.α B ) set independent asymptotes for the US-A and US-B associations; and B will compete with the US for association with A, and A will compete with the US for association with B (see Honey et al, 2020a , 2020b ).…”

“…For example, such individual variation is beyond the scope of general-process theories of associative learning (e.g., Rescorla & Wagner, 1972 ; Mackintosh, 1975 ; Pearce & Hall, 1980 ; Wagner, 1981 ) in which the relationship between the strength of an association and performance is held to be monotonic: How could a single acquired property (like associative strength; V) be manifest in distinct ways across a set of rats? We have recently presented a model, HeiDI, which offers a potential answer to this question ( Honey et al, 2020a , 2020b ). Before we present that answer, we should first describe results that provided an important impetus for the development of HeiDI.…”