The Dorsoventral Gradient Hypothesis: How Our Brains Might Balance Flexibility and Stability.

Have you ever wondered how your brain manages to react to split-second changes in your environment while simultaneously drawing on memories from decades ago? Or how you can recognize a specific dog while also understanding the concept of “dog” in general?

These seemingly contradictory abilities highlight a fundamental challenge for our brains: balancing immediate flexibility with long-term stability, and connecting specific instances to abstract concepts. A the recent meeting of the American Neuropsychiatric Association, we proposed a hypothesis for how this is achieved: a dorsoventral gradient that organizes neural processing from top (such as the parietal cortex) to bottom (including temporal lobes).

The Dorsoventral Plasticity Gradient Hypothesis

Imagine your brain as a mountainside with different layers. At the top (dorsal regions), the soil is loose and easily shifted—representing high neural plasticity. As you descend toward the bottom (ventral regions), the soil becomes increasingly packed and difficult to change—representing decreasing plasticity.

If present in the brain, this organizational principle could enables our brains to efficiently handle different types of information processing:

At the Top (Dorsal Regions):
– Connections form and dissolve easily
– Information processing happens in short time windows
– Networks form extensive, rapidly-changing associations
– Perfect for immediate perception and responses
– Information spreads over more physical space per unit time

At the Bottom (Ventral Regions):
– Connections form only with repetition or powerful signals
– Information integrates over longer time windows
– Networks form fewer, more stable associations
– Ideal for long-term, stable memory
– Requires less physical space as information is compressed

How Information Flows Through the Gradient
Going Down: From Real Time Experience to Long Term Memory, Instance to Abstraction
As information flows down the gradient, it undergoes a fascinating transformation. The mathematical formula that describes this process is:

Δwij = G(yj) · η · Σ[εj(t) · ai(t)]

In simpler terms, the ease of forming new connections (Δwij) decreases as we move ventrally (G(yj)), influenced by how strong the signal is (εj(t)) and how active the connected neurons are (ai(t)).

What does this mean in practice? Information with these qualities travels deeper into the ventral regions:

• Repeated many times (like words you’ve heard thousands of times)
• Emotionally powerful (like memories of significant life events)
• Strongly associated with other information (like concepts that connect to many other concepts)
• Meanwhile, weaker, single-occurrence, or loosely connected information doesn’t make it as far down. This natural filtering process helps create stable, abstract representations from the chaos of everyday experience.

Going Back Up: From Memory to Application
The process works bidirectionally. When we access a memory or concept:

• Stable ventral representations are activated
• They flow back up the gradient
• As they rise, they integrate with current contextual information
• This creates a rich blend of abstract knowledge and present-moment details
• This bidirectional flow explains how we can apply abstract knowledge to specific situations, recognize patterns across different contexts, and maintain a coherent understanding of the world despite constantly changing inputs.

Real-World Applications Across Brain Functions
This gradient organization explains multiple aspects of brain function:

Memory Systems
Working Memory (dorsal): Flexible, rapidly updating information we hold in mind temporarily
Semantic Memory (ventral): Stable knowledge about the world that persists over time

Language Processing
Dorsal Language Pathway: Handles phonological processing and speech production
Ventral Language Pathway: Manages meaning, concepts, and understanding

Visual Processing
Dorsal “Where/How” Pathway: Processes spatial relationships and guides action
Ventral “What” Pathway: Recognizes and categorizes objects

Motor Skills (Praxis)
Dorsal Regions: Handle real-time movement adjustments
Ventral Regions: Store “praxicons” (motor templates for familiar actions)

Evidence and Future Directions
This model makes several testable predictions:

• Different plasticity rates should be measurable between dorsal and ventral regions
• Time integration windows should vary across the gradient
• We should see different patterns of information flow during learning versus recall
• Advanced neuroimaging techniques like magnetoencephalography (MEG) or electrocorticography can test these predictions by measuring the timing and direction of neural activity.

Why This Matters: Insights into Brain Disorders
Understanding this gradient has clinical implications as well. We believe certain degenerative conditiosn like Alzheimer’s and frontotemporal dementias spread when diseased proteins are shared between neural connections over time. The hypothesis states that conditions in ventral regions should then spread slowly compared to those that spread in dorsal regions. This is what we see in semantic dementia, which spreads in ventral regions, and is the slowest progressing of these conditions.

The Elegant Efficiency of the Gradient
The dorsoventral gradient represents a remarkably efficient solution to fundamental computational challenges. By organizing processing along a continuum of plasticity, the brain achieves:

• Efficient real-time processing where needed
• Stable long-term storage where appropriate
• Natural filtering of information by importance
• Seamless integration of past knowledge with present experience

This organization helps explain not just how we remember and learn, but how we balance being creatures of habit with our capacity for spontaneity and creativity. Our brains are neither too rigid nor too chaotic—they exist in a carefully balanced state that allows us to be both grounded in what we know and responsive to what we encounter. The gradient model gives us a deeper appreciation for the elegant architecture that makes our minds simultaneously stable and flexible, capable of both specific details and grand abstractions—a true marvel of natural engineering. More study and review is needed, however, to verify this hypothesis. That’s how science works!