In a groundbreaking step towards understanding brain function, a team of biomedical engineers has uncovered a specialized biological mechanism that allows the brain’s thinking regions to reshape how sensory areas process identical visual information. This discovery opens new horizons for understanding brain dynamics and applying this knowledge to develop artificial intelligence models.
A New Mechanism for Understanding the Brain
Vision has long been considered a linear process, starting from the eyes and ending in the visual cortex where information is processed. However, new research suggests this process is more complex than previously thought. It turns out that the brain’s thinking regions can alter how visual information is processed based on context and required tasks.
This discovery was made possible by biologically inspired recurrent neural network models, which helped researchers isolate a specific mechanism known as “inhibition on inhibition.” This mechanism acts as a bridge between higher-level instructions and sensory perception, providing the brain with high cognitive flexibility.
Recurrent Neural Models as a Research Tool
Due to the limitations of techniques like fMRI in pinpointing individual cellular circuits, the team turned to building simple but biologically inspired recurrent neural models. These models include specialized clusters of excitatory and inhibitory neurons arranged in hierarchical structures. When trained to switch sorting rules, researchers discovered the role of inhibitory neurons that suppress other inhibitory cells, allowing information to pass from higher cognitive units to sensory inputs.
Live Experiments and Biological Confirmation
To verify the model’s accuracy, researchers recorded neural activity in the visual cortex of live mice. When similar inhibitory cells were disabled in living tissue, the cortex lost its ability to track task context, confirming the computational model’s validity. This highlights the repetitive nature of sensory functions in the brain and enhances our understanding of how the brain achieves this level of adaptation and flexibility.
Inspiration from Human Cases
The lead researcher drew inspiration from her work with patients lacking the hippocampus, a crucial brain structure for memory formation. Despite losing this critical part, these patients retained flexible cognitive skills, indicating that the brain’s primary sensory areas perform an adaptable surplus function.
Conclusion
This study opens new avenues for understanding how small brain circuits can provide insights into building more efficient artificial intelligence models. By studying mechanisms like inhibition on inhibition, engineers can design recurrent neural networks that are lean and adaptable, reducing the current models’ heavy reliance on energy and data. This represents a step towards AI models that surpass current ones in efficiency and adaptability.