MIT OCW Lecture: Causal Inference and Methods for Face Perception in the Brain (TMS, fMRI, and Animal Studies)

In this MIT OpenCourseWare lecture, the instructor surveys how scientists use multiple methods to study face perception in the brain. The talk contrasts behavioral data with functional imaging, invasive recording, and noninvasive brain stimulation, and it emphasizes the challenge of inferring causality from neural signals. A centerpiece is the discussion of two face processing hubs, the fusiform face area and the occipital face area, and how transcranial magnetic stimulation can test their causal roles in perception. The lecture also covers ethical issues in animal research, the power of animal studies to address deeper questions, and the practical design considerations that scientists face when planning experiments with minimal confounds.

• TMS as a tool to test causal involvement of face processing regions
• Distinctions between stimulus to brain causality and brain to behavior causality
• Value of animal studies for neural coding and development of face representations
• Principles of good experimental design and minimal pair contrasts

Overview

This lecture from MIT OpenCourseWare surveys how cognitive neuroscience addresses face perception using a spectrum of methods, from behavioral experiments to imaging and stimulation, and finally animal studies. The instructor emphasizes the distinction between observing neural responses and proving causal roles in perception, and he highlights the key brain regions involved in face processing, notably the fusiform face area (FFA) and the occipital face area (OFA).

Temporal and Spatial Resolution Across Methods

The talk reviews the tradeoffs between methods: functional MRI offers good spatial resolution but poor temporal precision and limited causal inference; scalp EEG/ERP provides excellent temporal data but limited spatial localization; invasive recordings give both but are restricted to clinical cases. Transcranial magnetic stimulation (TMS) is presented as a rare but powerful disruption method in normal subjects that can shed light on causality by transiently perturbing targeted cortex while subjects perform perceptual tasks.

TMS and Face Perception

The instructor explains how TMS is used to probe the OFA and the FFA. A classic finding is that stimulating the OFA around a brief time during face presentation impairs face perception more than stimulation at a control site, suggesting a causal role for OFA in early face processing. However, disrupting OFA does not reliably produce a conscious face percept, and the exact interpretation remains debated due to targeting limitations and the broader network dynamics. The timing of effects, such as a disruption window around 60 to 100 milliseconds post-stimulus, provides insight into when this region contributes to processing.

Animal Research and the Neural Code for Faces

The lecture shifts to animal studies, describing how Doris Tsao and Vinrich Freibled's work with monkeys has located a set of face patches and enabled direct recording from hundreds of neurons. This approach reveals the neural code for faces, how representations evolve over time, and the anatomical connectivity linking face patches. The speaker notes that such data allow questions about development, learning, and the influence of experience that are currently inaccessible in humans.

Ethics and Practical Considerations

Ethical considerations in animal research are addressed, including the regulatory frameworks in the United States, welfare concerns, and the balance between suffering and scientific gains. The discussion concludes with a pragmatic view of experimental design across human and animal work, stressing the importance of well constructed minimal pairs, baselines, run structure, and task design to isolate the mental processes under study.

Key Experimental Concepts

Terminology such as independent and dependent variables, hypotheses, predictions, confounds, contrasts, and the subtraction logic underpinning functional imaging contrasts is explained. The goal is to design experiments where observed differences can be attributed to specific cognitive processes while minimizing alternative explanations.

Takeaway

The field benefits from a diverse methodological toolkit that, when used thoughtfully, triangulates the causal role of brain regions in face perception. While human methods have limits, animal studies offer unparalleled opportunities to dissect neural codes and mechanisms across development and experience.