fbpx
Publications
COGNITIVE NEUROSCIENCE AS A MODEL FOR NEURAL SOFTWARE PATENT EXAMINATION

By Joseph S. Bird

INTRODUCTION

The Patent Office’s (“PTO”) internal classification system categorizes inventions mostly by structure, in the form of articles or products. This system is sufficient for many inventions but not for software performing functions analogous to the central nervous system (“CNS”). Neural software, also called neural computation or artificial intelligence (“AI”), is written to perform CNS functions and should be categorized using patent classifications to be developed based upon CNS functions which are observed in cognitive neuroscience research. These functions solve problems and thus enable survival of the animal in its environment.

Conventional software performs functions analogous to those of machines or real world objects. Neural software performs steps based upon analogies to mental steps.2 Conventional software patents already suffer from a high level of abstraction. Neural software patents inherit all the existing difficulties of other software patents and, in addition, suffer effects from an even higher level of abstraction.

Problem-solving categories would focus on the underlying functions involved rather than on the specific invention’s application. The problem-solving approach is already found in applicable case law governing patent validity.3 The first step in using new classifications would be to resolve whether a software invention is analogous either to a physical machine, implementing steps from rules, or to a process, like CNS functions.

To read this publication in its entirety click here.

RSS
Follow by Email
Facebook
Twitter