From TinyCog
Jump to: navigation, search


The TinyCog source code is open-source software licensed under the GNU Public License V3. You can download the latest version from the SourceForge download page or access the source on-line:

Please find below a copy of the current README file (2015-03-27):

For questions, suggestions and collaboration please Contact the authors directly or post to the SourceForge discussion forum.


The TinyCog code currently comes as a compressed archive in .tgz format. If you are on Windows, you can use [7-Zip] to decompress it. Just extract the contents of the .tgz in any sub-directory.

Please see the README below for how to execute TinyCog files.


	Minimalist Cognitive Architecture
	(c) 2015 Frank Bergmann (
	Licensed under GPL V3


About TinyCog

See for documentation.
TinyCog is designed as an easy to learn cognitive
architecture for academic use. It only consists of a few
thousand lines of code.

TinyCog License

All files are licensed under the GPL V3 unless otherwise
specified. Please contact the author for other licensing
terms. The GPL V3 explicitly disclaims any warranty for
the AI gone rogue :-)

TinyCog Training

Please contact for remote classes and
other TinyCog related training material.

Getting Started

TinyCog is written in SWI-Prolog, so you have to install
SWI-Prolog V6.6 or higher from
On Windows just double-click on any of the *.pl files below.
Each file contains a test predicate that will display status

TinyCog Modules

TinyCog consists of the following modules in order of
dependency from the most basic to the most advanced.

-	Auxillary predicates.

-	Storage back-end for scenes and objects:
		Uses tuples similar to the SWI-Prolog
		implementation of OWL (semantic Web).

-	"TinyDL" Description Logics:
		Minimalist DL system based on the ideas
		of FLEX (successor of BACK). Uses
		for it's TBox and ABox representation.

- Hanoi Towers test for +
		Plays hanoi using a built-in planner
		(not the full-blown planner.dl).

-	Simple Deterministic Planner:
		Builds plans (linked scenes) based on the
		"actions" available for agents.

- Hanoi Towers test for
		Defines a single "action" and simple
		proximity measure for hanoi towers.

- persgoals:	Persistent Goals for Agents:
		Defines a portfolio of goals for each
		agent that persist between actions and
		control the agent's behavior.

-	HTML5 Test Bed with "Hunter" domain:
		Defines "deer" and "hunter" agents on a
		1000 x 1000 playing field in order to
		test multi-agent plan coordination.