KembaraXtra – Psychology: Confusion Matrix

Definition

A confusion matrix is a table showing how often each stimulus, category, or response is correctly identified or mistaken for another. Rows usually represent the true categories, while columns represent the participant’s or model’s responses. The matrix summarizes both correct classifications and specific patterns of error.

Structure and Interpretation

Values along the main diagonal represent correct identifications, whereas off-diagonal values represent confusions. Frequent confusion between two stimuli suggests that they are perceptually or conceptually similar. Researchers may analyse these error patterns to infer the underlying dimensions that organize perception or classification.

Role in Psychology and Cognitive Science

Confusion matrices are used in studies of speech perception, face recognition, colour vision, memory, object identification, and sensory discrimination. For example, repeated confusion between red and green stimuli may indicate an abnormality in colour vision. The pattern of errors can reveal more than overall accuracy alone.

Applications in Machine Learning and Assessment

In machine learning, confusion matrices are used to calculate sensitivity, specificity, precision, recall, and overall accuracy. Similar methods are used in psychological testing and diagnostic classification. Interpretation must take category prevalence into account, because high accuracy can be misleading when one category is much more common than another.

Summary

A confusion matrix records correct and incorrect classifications across multiple categories. Its diagonal shows accurate responses, while other cells show specific errors. It is useful in perception research, diagnosis, and machine learning. The pattern of confusions provides insight into similarity and classification performance.