Symbolic processing computers were designed to address problems that conventional computing approaches could struggle to represent directly. Rather than working only with numerical values, symbolic systems manipulate symbols representing real-world objects, their properties and relationships. Texas Instruments’ Explorer family applied this approach to artificial intelligence (AI), while the microExplorer brought Explorer capabilities into a desktop system built around the Apple Macintosh II.
The design offers a useful historical example of how specialized processors, software environments and conventional computing resources can be combined. Its intended applications included expert systems for dynamic scheduling, equipment diagnostics, monitoring and control, configuration, and design—problem areas relevant to industrial automation and operations.
Why symbolic processing mattered
Many complex applications depend on rules, relationships and changing conditions, not just arithmetic. Symbolic processing provides a way to represent and work with that information. In an expert system, for example, an application can use encoded knowledge to help assess a situation or support a decision. The approach can also help developers build and revise software for problems whose requirements are difficult to define through conventional programming alone.
These capabilities require more than a programming language. They depend on an environment that supports symbolic software efficiently, including memory management and the handling of different data types. The Explorer family combined the Lisp language with software tools and hardware designed to support those operations.
Explorer architecture: software and hardware working together
The Explorer software environment included Common Lisp, Zetalisp extensions, the Flavors object-oriented programming system, and tools for prototyping and development. These components supported the creation of applications as well as their execution. The environment also offered reusable application building blocks, helping developers work at the level of the problem rather than implement every low-level function from scratch.
Hardware support was intended to reduce the processing overhead associated with symbolic operations. The Explorer architecture handled tags—bits associated with data words—alongside instruction execution. This allowed the system to support dynamic data types without requiring developers to declare every type in advance. It also supported automatic memory allocation and reclamation. TI’s Temporal Garbage Collection capability reclaimed memory incrementally during program execution.
At the center of the microExplorer processor was TI’s 32-bit Explorer Lisp microprocessor. The board included four megabytes of onboard memory, with expansion options described in the system specifications. It was designed to fit into an expansion slot in the Macintosh II, bringing specialized symbolic-processing hardware into a desktop computer.
Symbolic processing computers in a desktop system
The microExplorer paired two computing environments: the Explorer symbolic-processing system and the Macintosh II operating environment. The Macintosh provided a graphical interface, file system and access to its conventional software and communications products. Explorer applications could be started through the Macintosh interface and use its file system, rather than requiring users to manage separate disk storage environments.
The two processors could operate concurrently. A conventional application could handle tasks such as gathering data or performing numerical calculations, while an Explorer application supplied expert-system functions. Remote Procedure Call capabilities enabled applications on the two processors to exchange information, supporting integrated applications that used each environment for different parts of a task.
This division of work illustrates a practical systems principle: specialized computing does not necessarily need to replace general-purpose computing. In the microExplorer, the two were combined so that symbolic and conventional applications could run side by side and contribute to a larger workflow.
From application development to deployment
Explorer systems were positioned for both software development and application delivery. Larger Explorer systems provided development tools, while the microExplorer could be configured with a runtime version of the Explorer environment. That version omitted development tools to reduce software size, allowing applications developed on other Explorer family systems to be delivered on a smaller configuration.
Because the systems shared underlying software and hardware architectures, the stated aim was to avoid the porting difficulties that can arise when development and deployment platforms differ. Optional development software was also available for the microExplorer, so it was not restricted to running completed applications.
This development-to-delivery model addressed a common deployment consideration: the platform used to build an application may need different resources from the one used to run it. Separating development tools from runtime software offered a way to deliver applications without requiring every deployed system to include the full development environment.
Applications and implementation support
The Explorer family was presented as a platform for expert systems and other AI applications. The source material identifies several areas where such systems could be applied:
- Dynamic scheduling: supporting decisions as schedules and operating conditions change.
- Equipment diagnostics: applying encoded knowledge to help interpret equipment conditions.
- Monitoring and control: adding rule-based or knowledge-based capabilities to operational applications.
- Configuration and design: helping address problems involving multiple requirements and relationships.
- Software prototyping: developing and refining solutions for large, complex software systems.
TI also offered knowledge engineering services covering application and tool selection, feasibility demonstrations, prototyping, custom development, deployment, integration and maintenance. Training and technical support were part of the offering as well. These services recognized that an AI platform alone does not provide the application knowledge, integration work or ongoing support needed to put an expert system into operation.
Practical implications for automation teams
The microExplorer is a historical system, but its design highlights enduring considerations for industrial computing. First, match the computing architecture to the problem: applications involving rules and relationships may call for different tools from those focused primarily on numerical processing. Second, consider how specialized software will interact with established applications, data and user interfaces. Finally, plan for the full application lifecycle, including prototyping, deployment, integration, training and maintenance.
For automation engineers and technical decision-makers, the broader lesson is that useful AI depends on the relationship between the application, its software environment and the hardware that runs it. The Explorer family sought to bring these pieces together, while the microExplorer extended the approach into a desktop system capable of running symbolic and conventional applications concurrently.