Conjoint Contribution of Cognition and Emotion on the Control of Thought and Behavior Among Learners by Artificial Neural Networks

Abstract

The study developed new neural models for integrating cognition and emotion and assessing the conjoint contribution of the integrated neurons on the control of outputs (thoughts and behaviors) of learners. The neural model is then fed into different learning mechanisms, notably, error-minimization learning & Hebbian learning to determine the results of integration. Results revealed that (i) (i) (i) in general, the integration of cognition and emotion in a single neuron requires a greater number of training samples to achieve the same level of learning efficiency as in pure cognition models; (ii) a coupled neuron in which the emotion and cognition activities are inseparable leads to learning cycles and divergence in which the student never learns regardless of the learning mechanism used; and (iii) an uncoupled neuron on which the affective and cognitive neurons are integrated with separate synaptic weights can be trained but with more training samples than is normally required for cognitive learning. In the latter case, the restriction w_j,i>n_j,i implies that the teacher must begin by motivating the learner in order to compensate for a negative affective neuron and must consistently do so throughout the lesson in order for the neural synapses to be trained, i.e., for the learners to understand the subject matter. In more practical terms, the results also indicate that in terms of pedagogy, more teacher-learner contacts will be required to achieve acceptable learning efficiency levels. Further, at each contact, the synaptic weights corresponding to positive emotions will need to be increased while the synapses corresponding to negative emotions will need to be decreased.