Abstract
A great deal is known about the response properties of single neurons processing sensory information. In contrast, less is understood about the collective characteristics of networks of neurons that may underlie sensory capacities of animals. We used functional magnetic resonance imaging to study the emergent properties of populations of neurons processing motion across different brain areas. Using a visual adaptation paradigm, we localized a distributed network of visual areas that process information about the direction of motion as expected from single-cell recording studies. However, we found an apparent discrepancy between the directional signals in certain visual areas as measured with blood oxygenation level-dependent imaging compared with an estimate based on the spiking of single neurons. We propose a hypothesis that may account for this difference based on the postulate that neuronal selectivity is a function of the state of adaptation. Consequently, neurons classically thought to lack information about certain attributes of the visual scene may nevertheless receive and process this information. We further hypothesize that this adaptation-dependent selectivity may arise from intra- or inter-area cellular connections, such as feedback from higher areas. This network property may be a universal principle the computational goal of which is to enhance the ability of neurons in earlier visual areas to adapt to statistical regularities of the input and therefore increase their sensitivity to detect changes along these stimulus dimensions.
Introduction
For decades, systems neuroscience has relied heavily on single-unit electrophysiology to characterize the response properties of individual neurons. While this has provided deep insights into areas like V1 (primary visual cortex) and MT (middle temporal area), it offers a limited view of how these regions function as an integrated population. A significant challenge in studying population-level direction selectivity with fMRI is spatial resolution; neurons with opposing direction preferences are typically intermingled within a single voxel, causing their signals to cancel out in standard subtraction paradigms.
This study by the Logothetis lab (2001) addresses this limitation by employing a visual adaptation paradigm. By "fatiguing" specific neuronal populations, the researchers were able to uncover hidden directional signals and explore the dynamic nature of neuronal selectivity across the macaque visual hierarchy.
Methods
The researchers performed high-field (4.7T) fMRI on four anesthetized macaque monkeys (Macaca mulatta). To isolate direction-selective responses, they utilized a visual adaptation paradigm:
- Stimulus: A high-contrast rotating checkerboard pattern (Figure 1).
- Adaptation Phase: The stimulus rotated in a single direction for 200 seconds, leading to a gradual decline in the BOLD signal as direction-selective neurons adapted.
- Test Phase (Rebound): The direction of rotation was abruptly reversed. If the region contained direction-selective neurons, the previously un-adapted population would respond, creating a "rebound" peak in the BOLD signal.
- Analysis: A BOLD directionality index was calculated as the ratio of the rebound response to the initial response. This fMRI-derived metric was then compared to predictions based on historical single-cell spiking data.
Results
The adaptation paradigm successfully mapped a distributed motion-processing network, including areas V1, V2, V3, V3A, V4, and MT.
The BOLD-Spiking Discrepancy
The most striking finding was in area V1. Electrophysiological data suggests only ~20% of V1 neurons are direction-selective, which predicts a BOLD directionality index of approximately 0.11. However, the observed fMRI index was 0.33—three times higher than expected. Conversely, area MT showed a high index (0.84), consistent with its known role as a motion-processing hub. Area V4 also showed a surprisingly high index of 1.0, despite its traditional association with color and shape processing.
Temporal Dynamics of Adaptation
The study revealed distinct adaptation profiles across the hierarchy. V1 exhibited "slow" adaptation (time constant $\tau \approx 102$ s), whereas higher-order areas like MT and V2/V3 adapted much more rapidly ($\tau \approx 28-37$ s). This suggests that higher areas are more efficient at filtering out predictable sensory input.
Discussion
The discrepancy between spiking output and BOLD signal in V1 suggests that BOLD imaging may be more sensitive to synaptic inputs (subthreshold activity) than to the final spiking output of neurons. The authors propose that "non-directional" neurons in V1 may still receive and process directional synaptic information, which becomes evident under adaptation.
Furthermore, the results suggest that neuronal selectivity is not a static property but a dynamic one influenced by the state of adaptation and potentially by feedback from higher-order areas. This mechanism may allow the visual system to remain highly sensitive to changes in environmental statistics, enhancing the detection of novel motion signals. While anesthesia remains a potential confound for feedback-dependent processes, this study laid the groundwork for understanding the complex relationship between local neuronal activity and global hemodynamic responses.
Summary updated