Of the Artistic Nude and Technological Behaviorism
Leon Harmon and the First Steps towards Neuromorphic Hardware
Introduction: When Art Meets Neuroscience
In an unexpected fusion of aesthetics and engineering, the pioneering work of 1960s researcher Leon Harmon forever bridged the world of artistic imagery and neurological computation. What could the artistic nude possibly have to do with sophisticated brain-inspired technology? The answer lies in Harmon's revolutionary investigations into visual perception using early "neuromorphic" devices—simple electronic circuits that mimicked the behavior of neural systems. At a time when computers filled entire rooms and artificial intelligence was in its infancy, Harmon conceived of elegant, low-power devices that could process information much like biological brains do 1 .
This research represents a crucial early chapter in what we now call neuromorphic engineering—the creation of devices that emulate the neural structure of the brain.
By understanding these first steps toward brain-inspired computing, we can better appreciate today's ambitious projects like the Human Brain Project and the Brain Initiative, which continue Harmon's quest to reverse-engineer the most complex object in the known universe 1 .
Neuromorphic Engineering
The interdisciplinary field that takes inspiration from the structure and function of the brain to design efficient computing systems.
Art-Science Intersection
Harmon's work demonstrates how artistic imagery can reveal fundamental principles of neural processing and perception.
What is Neuromorphism? The Brain as Machine
Neuromorphic technology represents a fundamental shift from traditional computing. Rather than processing information through sequential instructions in a central processing unit, neuromorphic systems mimic the brain's architecture—with many simple units (artificial neurons) working in parallel, communicating through connections (artificial synapses) of varying strengths. The term "neuromorphic" literally means "taking the form of nerves," and these systems capture both the structural and functional principles of biological neural networks 1 .
Traditional Computing
- Sequential processing
- High power consumption
- Fixed architecture
- Centralized control
Neuromorphic Computing
- Parallel processing
- Low power consumption
- Adaptive architecture
- Distributed processing
Key Advantages of Neuromorphic Systems
Lower Power Consumption
Thousands of times more efficient than conventional processors
Higher Computational Speed
Parallel processing enables simultaneous operations
Real-time Processing
Ideal for sensory processing requiring immediate response
Compact Physical Volume
Efficient design enables complex computation in small spaces
The conceptual foundation for neuromorphism dates back to the 1940s, when researchers like Warren McCulloch began seriously proposing that the brain could be understood as a computing machine 1 . What Harmon contributed in the 1960s was one of the first practical implementations of this theoretical possibility—the "neuromime," an electronic circuit that could simulate the basic behavior of neurons.
The Lincoln Experiment: When a Portrait Revealed Neural Principles
In 1973, Leon Harmon created what would become one of his most famous demonstrations—the quantized Lincoln portrait. This experiment brilliantly connected artistic imagery with the principles of neural processing, offering a striking visual representation of how our visual system might encode and recognize images 1 .
Methodology: From Image to Electrical Impulses
Harmon's experimental approach followed several carefully designed steps:
Image Selection and Simplification
Harmon began with a recognizable image—a portrait of Abraham Lincoln—and converted it into a simplified grayscale representation. This reduction removed unnecessary detail while maintaining essential features needed for recognition.
Brightness Quantization
Each area of the image was assigned one of twelve discrete brightness levels, transforming the continuous-tone photograph into a pattern of discrete values. This step mirrored how biological visual systems have limited resolution and sensitivity ranges.
Neuromime Processing
The quantized image was processed through Harmon's electronic neuron analogs. These circuits simulated key properties of biological neurons, including non-linear response characteristics, basic learning capabilities, and signal transmission similar to neural firing patterns.
Output Analysis
The processed output was analyzed both for its visual properties and for what it revealed about underlying neural processing mechanisms 1 .
Brightness Quantization Levels
| Brightness Level | Relative Intensity | Visual Appearance | Role in Recognition |
|---|---|---|---|
| Level 1 (Darkest) | 0-8% | Near black | Defines shadow regions |
| Level 3 | 17-25% | Dark gray | Contour definition |
| Level 6 | 42-50% | Medium gray | Mid-tone features |
| Level 9 | 67-75% | Light gray | Highlight areas |
| Level 12 (Lightest) | 92-100% | Near white | Specular highlights |
Results and Analysis: The Science of Recognition
When viewers were presented with Harmon's processed Lincoln image, they could still recognize the subject despite the drastic simplification. This demonstrated that human visual perception relies on specific key features rather than detailed pixel-by-pixel analysis. The experiment revealed several fundamental principles of visual processing:
Feature Extraction
Precedes recognition; our visual system detects edges, contours, and patterns before identifying what we're seeing
Pattern Completion
The brain fills in missing information when presented with partial data
Non-linear Processing
Enables robust perception across various transformations and degradations
The quantized Lincoln thus served as both scientific evidence and powerful demonstration—revealing the core principles our brains use to make sense of visual information 1 .
The Scientist's Toolkit: Components of Neuromorphic Research
The field of neuromorphic engineering relies on specialized components and approaches that enable researchers to emulate neural functions. While modern tools have evolved significantly since Harmon's era, the fundamental principles remain remarkably consistent.
| Component/Concept | Function | Biological Analog |
|---|---|---|
| Neuromime Circuits | Basic building blocks that simulate neuron firing behavior | Biological neurons with activation thresholds |
| Synaptic Weight Modules | Adjustable connections that determine signal strength between artificial neurons | Chemical synapses between neurons |
| Pulse Coding Systems | Encode information as timed electrical spikes rather than continuous values | Action potential firing in neural systems |
| Adaptive Plasticity Mechanisms | Allow the system to modify connection strengths based on experience | Synaptic plasticity in biological brains |
| Lateral Inhibition Networks | Enable contrast enhancement and feature detection in sensory processing | Inhibitory interneurons in sensory cortices |
Harmon's particular innovation was implementing functionally equivalent rather than biologically identical components. His neuromimes didn't precisely replicate the complex biophysics of real neurons but captured their essential input-output behavior in a form that engineers could readily work with 1 .
This approach, sometimes called "technological behaviorism," focuses on replicating the external behavior of neural systems rather than their precise internal mechanics—a pragmatic strategy that has enabled rapid progress in neuromorphic engineering.
This toolkit has evolved dramatically since the 1960s. Where Harmon worked with discrete electronic components, today's researchers create silicon implementations of pulse-coded neural networks that can contain millions of artificial neurons on a single chip 4 . The principles, however, remain recognizably similar to those Harmon pioneered.
Evolution of Neuromorphic System Complexity
Technological Behaviorism: The Philosophy Behind the Hardware
Harmon's approach was guided by what might be called technological behaviorism—the principle that for engineering purposes, what matters is functional equivalence rather than detailed biological accuracy 1 . This perspective asks: does the artificial system process information in a way that produces similar outputs to biological systems, even if its internal mechanisms differ?
Functional Equivalence
Focuses on replicating the input-output relationships of biological systems rather than their internal mechanics.
Pragmatic Engineering
Emphasizes practical implementation with available technology over theoretical perfection.
This philosophy represented a pragmatic middle ground between abstract mathematical models of neural networks and attempts to create biologically precise simulations. By focusing on input-output relationships rather than biochemical details, Harmon and his contemporaries could build working systems with the technology available to them.
The technological behaviorism approach has proven remarkably enduring in neuromorphic engineering. Modern brain-inspired chips still prioritize functional equivalence over biological accuracy, while incorporating more detailed biological constraints as semiconductor technology allows.
Legacy and Impact: From Neuromimes to Modern Neuroscience
Leon Harmon's work established foundational principles that continue to guide neuromorphic research today. His quantized images demonstrated that simplified neural models could capture essential aspects of complex perceptual processes—a revelation that influenced both neuroscience and artificial intelligence.
Evolution of Neuromorphic Technologies
| Era | Key Technologies | Scale/Complexity | Primary Applications |
|---|---|---|---|
| 1960s (Harmon) | Discrete neuromime circuits | Dozens of neurons | Basic perception models, visual processing |
| 1980s-1990s | Custom VLSI neural chips | Hundreds to thousands of neurons | Signal processing, early robotics |
| 2000-2010 | Field-programmable analog arrays | Thousands to millions of neurons | Sensory systems, motor control |
| 2010-Present | Large-scale neuromorphic systems (SpiNNaker, Loihi) | Millions to billions of neurons | Brain simulation, AI, real-time processing |
Key Developments Since Harmon's Era
While Harmon's models emphasized functional equivalence, subsequent research has incorporated more detailed biological mechanisms, including different neuron types and complex synaptic plasticity rules.
The transition from discrete circuits to integrated systems culminated in silicon implementations of pulse-coded neural networks, which allowed for unprecedented complexity and scale 4 .
Harmon's work exemplified the interdisciplinary nature of evolutionary approaches to understanding human behavior and cognition—bridging engineering, neuroscience, and psychology .
Today's major brain projects—the American Brain Initiative and European Human Brain Project—represent massive, coordinated efforts to achieve what Harmon began on a small scale: reverse-engineering the brain's computational principles 1 .
Timeline of Neuromorphic Technology Development
Conclusion: The Enduring Bridge Between Art and Science
Leon Harmon's work with neuromimes and his striking quantized images established a crucial bridge between artistic representation and neurological function that remains relevant more than half a century later. His research demonstrated that simple electronic systems could capture essential aspects of how our brains process visual information, while his technological behaviorism approach provided a practical philosophy for building brain-inspired technology.
Today, as neuromorphic engineering enters its most exciting phase—with brain-scale simulations becoming feasible and neuromorphic chips enabling new forms of efficient computation—we can appreciate Harmon's contributions as prescient first steps toward understanding and emulating the brain's remarkable capabilities.
The fusion of art and technology in his Lincoln portrait perfectly symbolizes the interdisciplinary nature of this quest, reminding us that understanding the brain's inner workings requires both scientific precision and artistic insight.
As contemporary projects like the Human Brain Project continue to advance Harmon's vision, the fundamental principles he established—functional equivalence, pragmatic engineering, and the value of simple models—continue to guide researchers in their pursuit of one of science's ultimate goals: understanding and recreating the computational power of the human brain.