Acquired equivalence associative learning in GTC epileptic patients: experimental and computational study

Abstract

Previous cognitive behavioral studies based on Acquired Equivalence Associative learning Task (AEALT) showed a strong relation between hippocampus and basal ganglia in associative learning. However, experimental behavioral studies of patients with Generalized Tonic Clonic (GTC) epilepsy remained sparse. The aim of the present study is to integrate a classical behavioral cognitive analysis with a computational model approach to investigate cognitive associative learning impairments in patients with GTC epilepsy. We measured the accuracy of associative learning response performance in five GTC epileptic patients and five control subjects by using AEALT, all subjects were matched in age and gender. We ran the task using E-Prime, a neuropsychological software program, and SPSS for data statistical analysis. We tested whether GTC epileptic patients would have different learning performance than normal subjects, based on the degree and the location of impairment either in basal ganglia and/or hippocampus. With the number of patients that was available, our behavioral analysis showed no remarkable differences in learning performance of GTC patients as compared to their control subjects, both in the transfer and acquisition phases. In parallel, our simulation results confirmed strong connection and interaction between hippocampus and basal ganglia in our GTC and their control subjects. Nevertheless, the differences in neural firing rate of the connectionist model and weight update of basal ganglia were not significantly different between GTC and control subjects. Therefore, the behavioral analysis and the simulation data provided the same result, thus indicating that the computational model is likely to predict cognitive outcomes.

Keywords: acquired equivalence associative learning task; basal ganglia; cognitive impairment; connectionist model; generalized tonic clonic epilepsy; hippocampus.

Figure 1

It shows Acquired Equivalence Associative learning Task (AEALT); this task is (adapted from Moustafa et al., ; Myers et al., ; Herzallah et al., ; Moustafa and Gluck, 2011). (A) Represents the phases of AEALT; acquisition and transfer phases (see task description, Methodology Section), whilst (B) is a screen snapshot represents experimental trials in an early stage of acquisition phase which is considered as shaping and training stage. In this latter stage, the stimulus is represented as a human face appearing on the screen and subsequently, the subject responds by choosing one of the colored fish either by clicking left or right, only then, the correct feedback is given (see task description, Methodology Section).

Figure 2

It represents structural description of the connectionist model, see text for illustration, Methodology Section.

Figure 3

This figure represents the data of our subjects which considered as an input for our computational model. Panel (A) shows the average of the accurate response in associative learning in GTC epileptic patients and their healthy controls during both phases of AEALT; acquisition and transfer phases of whilst the associated table to (A) represents the mean and the standard deviation of the average accuracy responses. (B) Represents the average of block trials which were performed by the GTC epileptic patients and their healthy controls during acquisition and transfer phase, whereas these values, which are represented in the associated table to (B), were required as input for our simulated model. (C) Indicates the average of block trials which were performed by the GTC epileptic patients and their healthy controls during the stages of shaping and equivalence training, and in the new consequent one, which resemble the acquisition phase, whereas these values, which are shown in the adjacent table to (C), were required as inputs for our simulated model.

Figure 4

This figure represents the average of the neural firing rates in response to the connection between basal ganglia and hippocampus when representing different values of hippocampus input indicating hippocampal strength. Panel (A) shows the average of neural firing rates when hippocampal input equal to zero, which triggers the neural firing rate relates to GTC epileptic patients and their controls as an outcome of our simulated model. The X axis represents the learning rate of basal ganglia, as basal ganglia input, and hippocampus input, as hippocampus strength, whereas a zero value of hippocampus input refers to its absence whilst Y axis shows the proportion of the neural firing rate which is in approximation below 55% for both subjects. In GTC-epileptic patients, this firing rate is lower than the controls except when the values of learning rate parameters of the basal ganglia are 0.2 and 0.8 and this rate tends to increase to be higher in GTC than in controls when the learning rate value is 0.9. (B) Represents the evaluation of weight update in basal ganglia module when hippocampus strength is absent, as in (A), whereas hippocampus input is equal to zero. This evaluation is represented in terms of percentages referring to GTC epileptic patients and their controls as an outcome of our simulated model. Notably, the percentage of weight update is still below 55% for both GTC epileptic patients and controls, whereas the evaluation of weight update of the basal ganglia relates to the action of the direct actor either right or left. In GTC epileptic patients, the evaluation of weight update of basal ganglia is higher as compared to controls when the learning rate value of basal ganglia is below 0.6. Above 0.6, this evaluation starts to decline to be lower than that of control. Additionally, this evaluation of the weight update is in synchronization with that of controls when the learning rate of basal ganglia is around 0.9. Panel (C) shows the average of neural firing rates when hippocampal input is above zero, taking the values of 1, 2, 3, and 4, whereas each value resemble a different state of hippocampus strength which is represented mathematically as input values. Controversially to (A), Hippocampus inputs with values higher than zero refers to the presence of hippocampus with differential strengths. The learning rate values of the basal ganglia were critical for modulating the neural firing rates produced by stabilizing it. However, as long as they increase, the firing rate activities remained comprised between 68 and 81%. (D) Represents the evaluation of weight update in basal ganglia module when hippocampus strength, input, is above zero, taking the same values mentioned in (C). In general, the evaluation of weight update of the basal ganglia increased rapidly when the hippocampus strength was large (values 3 and 4) while it slowly increased when the hippocampus strength was less (values 1 and 2).