Mechanisms of alcohol influence on fear conditioning: A computational model

Abstract

Background: A connection between stress-related illnesses and alcohol use disorders is extensively documented. Fear conditioning is a standard procedure used to study stress learning and links it to the activation of amygdala circuitry. However, the connection between the changes in amygdala circuitry and function induced by alcohol and fear conditioning is not well established.

Methods: We introduce a computational model to test the mechanistic relationship between amygdala functional and circuit adaptations during fear conditioning and the impact of acute vs. repeated alcohol exposure. Using firing rate formalism, the model generates electrophysiological and behavioral responses in fear conditioning protocols via plasticity of amygdala inputs. The influence of alcohol is modeled by accounting for known modulation of connections within amygdala circuits, which consequently affect plasticity. Thus, the model connects the electrophysiological and behavioral experiments. We hypothesize that alterations within amygdala circuitry produced by alcohol cause abnormal plasticity of amygdala inputs such that fear extinction is slower to achieve and less robust.

Results: In accordance with prior experimental results, both acute and prior repeated alcohol decrease the speed and robustness of fear extinction in our simulations. The model predicts that, first, the delay in fear extinction caused by alcohol is mostly induced by greater activation of the basolateral amygdala (BLA) after fear acquisition due to alcohol-induced modulation of synaptic weights. Second, both acute and prior repeated alcohol shift the amygdala network away from the robust extinction regime by inhibiting activity in the central amygdala (CeA). Third, our model predicts that fear memories formed during acute or after chronic alcohol are more connected to the context.

Conclusions: The model suggests how circuit changes induced by alcohol may affect fear behaviors and provides a framework for investigating the involvement of multiple neuromodulators in this neuroadaptive process.

Keywords: amygdala circuit; computational model; electrophysiological activity; fear acquisition; fear extinction.

FIGURE 1

Diagram of the amygdala circuitry implemented in the model. LA, lateral amygdala, BAe and BAf are extinction and fear groups of the basal amygdala. CeAOff and CeAOn are pro‐extinction and pro‐fear neural groups in the central amygdala, respectively.

FIGURE 2

Design of the simulated trial structure. (A) A simulated session consists of 25 fear acquisition trials followed by 35 fear extinction trials. (B) Each individual trial is simulated for 15 sec. followed by an intertrial interval. For the acquisition trials (left), a CS represented by a thalamic input to the LA and context represented by a hippocampal input (Hipp) to the BAf are on. For the extinction trials (right), the CS input is active again, but a mPFC group projecting to BAe is on instead of the hippocampal input. Each acquisition trial ends with a shock (UC), whereas in the extinction trials, the shock is omitted.

FIGURE 3

Dynamics of amygdala activity and synaptic inputs during fear acquisition and extinction. Acquisition trials are 0–14. Extinction trials are 25–59. Trials 15–24 represent a pause or “home cage rest” (shaded). The vertical dashed line marks the first trial after the pause representing fear recall in a new context. (A) Average activity levels of LA, BAf, and BAe. (B) Average activity levels of CeAOn and CeAOff. (C) Weights of synaptic inputs to LA, BAf, and BAe. (D) Exemplary time dependence of the activity levels throughout the first 20 trials for the BLA neural groups. The averaging in (A) and (B) is done over the first 15 sec of each trial, whereas the remaining 7.5 s is an intertrial interval.

FIGURE 4

Number of trials to criteria (i.e., fear response in acquisition and the absence of it in extinction) for the control (Naïve), acute and chronic alcohol conditions. The bars show the mean, and the brackets show the standard deviation over 100 sessions.

FIGURE 5

Effects of acute alcohol on amygdala activity during fear acquisition and extinction. Notation is the same throughout the panels as in Figure 3. Parameter values changed from control are in Table 2.

FIGURE 6

Effects of chronic alcohol on amygdala activity during fear acquisition and extinction. The notation is the same as in Figure 3. Parameter values changed from the control are in Table 3.

FIGURE 7

Within‐trial analysis of model dynamics in naïve (A), acute (B) and chronic (C) alcohol conditions. The system is reduced to two variables: Activation of CeAOn and CeAOff neurons. The red and blue curves are the nullclines of the corresponding equations. Their intersections are equilibrium states. The gray is a noisy trajectory showing robust activation of the CeAOff neurons in the naïve state (A), but their much weaker and less robust activation in acute (B) and especially in the chronic alcohol case (C).