Deep learning: A primer for psychologists

Area of Science:

• Psychology
• Computer Science
• Data Science

Background:

• Deep learning (DL) has transformed fields like computer vision and natural language processing.
• DL is underutilized in psychological research for predictive modeling.
• Researchers need accessible introductions to DL for psychological applications.

Purpose of the Study:

• To introduce deep learning concepts and models to psychologists with a linear regression background.
• To demonstrate the application of DL for predictive modeling in psychological research.
• To provide practical examples using R code for common psychological data types.

Main Methods:

• Overview of DL principles.
• Explanation of three core DL models: feedforward neural networks (FNNs), recurrent neural networks (RNNs), and convolutional neural networks (CNNs).
• Illustrative toy examples with R code.

Main Results:

• Demonstrated how FNNs, RNNs, and CNNs generalize linear regression.
• Provided practical R code for applying DL models to psychological data.
• Showcased DL's potential for prediction-focused research questions in psychology.

Conclusions:

• Deep learning offers significant benefits for predictive modeling in psychology.
• Basic DL models can be understood and applied by researchers familiar with linear regression.
• This work facilitates the adoption of DL techniques in psychological research for enhanced data analysis and prediction.