# Cognitive Architectures: how do you perform qualitative and quantitative comparisons?

I couldn't find a sub stackexchange for artificial intelligence, but I think robotics comes close, and so I'm posting here.

I recently saw TED talks on AI and the Google car, with these being the most interesting to me:

The third one led me to the 'criticism' section (labeled Analysis on that wiki article, though it certainly at least partially reads as a criticism section as well) of Kurzweil 'theory' of the brain, namely "Pattern Recognition Theory of Mind" (PRTM). After some link surfing on the people who have performed analysis of PRTM and their respective academic contributions, I came to learn about Cognitive Architecture:

"A cognitive architecture can refer to a theory about the structure of the human mind. One of the main goals of a cognitive architecture is to summarize the various results of cognitive psychology in a comprehensive computer model. However, the results need to be in a formalized form so far that they can be the basis of a computer program. By combining the individual results are so for a comprehensive theory of cognition and the other a commercially usable model arise. Successful cognitive architectures include ACT-R (Adaptive Control of Thought, ACT), SOAR and OpenCog."

It appears that there are several interesting architectures, including the 3 mentioned above. I read a bit about ACT-R, SOAR, OpenCog, DUAL, CHREST, and CLARION. The list is not comprehensive. It also appears that there are two main types of such architectures: Connectionism and Symbolic.

Though I have many questions, my main question is this: What are some quantitative metrics and qualitative properties to measure and compare between the two architecture types?

Other questions

• Can all architectures be categorized as one, the other, or some combination of the two, or is there a third, fourth, etc?
• How are two main types alike? How are they different?
• What are some recommended further readings on this topic.
• What centres and organizations are leading development in this?
• What are some of the computer programming languages, related skill-sets, and cross-domain knowledge set utilized in R&D and product offerings of such systems?
• First, OP stands for "Original Poster" - the person that asked the question. This is you. Second, questions that should be asked here are questions that are closed-ended, in that they should have 1 correct answer. The only question you have asked that meets this criteria is 1.3 - Is there a way to benchmark architecture types. I would ask that you edit your question to only ask that one, and take all of the other questions to the Robotics chat room. Jun 9, 2015 at 13:23
• Welcome to robotics Saad Shakil, but I'm afraid that open ended questions like this really aren't a good fit for a stack exchange site. We prefer practical, answerable questions based on actual problems that you face. Take a look at How to Ask and tour for more information on how stack exchange works. Jun 12, 2015 at 14:16
• I've asked the question that meets the criteria in a less open-ended manner. Does it qualify to be taken off hold now? I believe it does 'inspire answers that explain “why” and “how”', it will likely 'tend to have long, not short, answer', I certainly think it has a 'constructive, fair, and impartial tone', I look fwd to knowing 'experiences over opinions', certainly all of it 'backed up with facts and references', and certainly 'more than just mindless social fun'. Jun 13, 2015 at 7:07
• I still feel that it is too broad, as it probably falls under the category of needing a book to fully explain, but I'll reopen it as an experiment, to see if anyone can have a go at answering it. Jun 15, 2015 at 13:05
• There are more than two architecture types. Are you looking for benchmarking any two architecture types, or are you looking to compare approaches? You linked to connectionism, can you link to symbolic as well? Jun 15, 2015 at 18:18