- Email: [email protected]
Thoughts on bands of action
Available online at www.sciencedirect.com Available online at www.sciencedirect.com
ScienceDirect ScienceDirect
Procediaonline Computer 00 (2019) 000–000 Available at Science www.sciencedirect.com Procedia Computer Science 00 (2019) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
ScienceDirect Procedia Computer Science 145 (2018) 710–716
Postproceedings of the 9th Annual International Conference on Biologically Inspired Cognitive Postproceedings of the 9thBICA Annual International Conference Inspired Cognitive Architectures, 2018 (Ninth Annual MeetingonofBiologically the BICA Society) Architectures, BICA 2018 (Ninth Annual Meeting of the BICA Society)
Thoughts on bands of action Thoughts on bands of action Philip C. Jackson, Jr. * Philip C. Jackson, Jr. * 0F
0F
TalaMind LLC, PMB #363, 55 E. Long Lake Rd., Troy, MI 48085, U.S.A. TalaMind LLC, PMB #363, 55 E. Long Lake Rd., Troy, MI 48085, U.S.A.
Abstract Abstract This paper considers the ‘bands of action’ discussed by Newell in 1990. Reasons are given for a larger range of intelligent processing happen in cognitive and rational bandsbythan Newell Different reasons proposed This paper to considers thethe‘bands of action’ discussed Newell in described. 1990. Reasons are given forarea given larger from rangethose of intelligent by Newell for considering social band exist and be than distinct fromdescribed. the rationalDifferent band. Although suggested ~10proposed minutes processing to happen in thethe cognitive and to rational bands Newell reasons Newell are given from those might be a for minimal timescale for significant actions social from band,the reasons areband. givenAlthough for minimal timescales of seconds. The by Newell considering the social band to exist andinbethedistinct rational Newell suggested ~10 minutes importance of the social band in to create theincognitive and rational bands discussed. The impact of modern technology might be a minimal timescale forhelping significant actions the social band, reasons areisgiven for minimal timescales of seconds. The on the socialofband is described. Some potential therational Common Model of Cognition are suggested. A technology question is importance the social band in helping to createimplications the cognitivefor and bands is discussed. The impact of modern asked timescale of inspirations. on theabout socialthe band is described. Some potential implications for the Common Model of Cognition are suggested. A question is asked about the timescale of inspirations. © 2019 The Authors. Published by Elsevier B.V. © 2018 The Authors. by Elsevier B.V. This is an open accessPublished article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the the CC scientific committee of the 9th Annual International Conference on Biologically Inspired This is an open access article under BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Architectures. Architectures. Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Cognitive Architectures. Keywords: cognition, bands of action, rational band, social band, natural language, inspirations Keywords: cognition, bands of action, rational band, social band, natural language, inspirations
1. Newell’s ‘bands of action’ 1. Newell’s ‘bands of action’ In considering human cognition, Newell [8] discusses different time scales at which human action can occur. He In considering [8] discusses scales at whichbands human action can occur. and He groups these timehuman scalescognition, into the Newell following “bands ofdifferent action”,time saying different have phenomenal groups these time scales into the following “bands of action”, saying different bands have phenomenal and
* Corresponding author. address:author. [email protected] * E-mail Corresponding URL: http://www.talamind.com E-mail address: [email protected] URL: http://www.talamind.com 1877-0509 © 2019 The Authors. Published by Elsevier B.V. This is an open access under the CC by BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) 1877-0509 © 2019 The article Authors. Published Elsevier B.V. Peer-review under responsibility of the committee of the(https://creativecommons.org/licenses/by-nc-nd/4.0/) 9th Annual International Conference on Biologically Inspired Cognitive This is an open access article under thescientific CC BY-NC-ND license Architectures. Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Architectures.
1877-0509 © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Architectures. 10.1016/j.procs.2018.11.049
2
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
711
theoretical differences: Biological band Organelle (10-4 secs) Neuron (10-3 secs) Neural circuit (10-2 secs) Cognitive band Deliberate act (10-1 secs) Cognitive operations (1 sec) Unit task (10 sec) Rational band Tasks ranging from minutes to hours Social band Actions ranging from days to months Historical band† Actions ranging from years to millennia Evolutionary band Actions ranging from 104 to 106 years Except for the evolutionary band, these bands of action all correspond to real, physical events that happen due to processes within human brains. The evolutionary band involves physical events at the level of the human genome.‡ 2. The cognitive and rational bands After discussing operations by neural circuits, Newell [8] observed there is very little time available in the biological band to achieve cognitive behavior (p.129). To show this, he discussed the processing that must occur in a simple human interaction when one person asks “Please pass the salt” and another person responds in about a second. In the one second for a person to hear and understand "Please pass the salt" and then reach for the salt, there is only time to perform about 10 ‘deliberate acts’ of cognition. The one second includes speech recognition, natural language understanding, deciding whether to pass the salt, and then initiating an act to pass the salt (which could involve remembering where the salt was, looking for the salt, etc.). Newell noted it can also involve deciding whether the request to pass the salt was polite or impolite, and it might take longer to respond if the request was impolite. Newell observes that one consequence of the time scales required for biological processing is that the human computational system must be massively parallel to accomplish complex tasks in the timescale of the cognitive band. As an example of more complex cognitive processes requiring about 10 seconds for a unit task, Newell discusses playing ‘rapid-transit chess’, with rules allowing only 10 seconds per move. Newell (p.144) gives other examples of cognitive tasks, e.g. solving the CROSS+ROADS problem at about 2 seconds and solving a logic problem at about 8 seconds. Newell [8] writes (p.150) that as more time is available for solving problems, the system will have more time to find solutions using knowledge it already has. It may be described as a ‘knowledge-level system’, operating in the rational band. In the rational band an agent uses knowledge to achieve goals.
†
Figure 3-3 of [8] showed the first four bands. Figure 3-14 added the historical and evolutionary bands. Except where noted otherwise, page and figure numbers cited in this paper are relative to reference [8]. ‡ With current technologies, human actions may affect evolution in much smaller timescales.
712
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
3
In another paper [4], I give reasons why Newell’s definition of the knowledge level [7] is problematical. I discuss theoretical and practical reasons to recognize the existence of an ‘intelligence level’ corresponding to human-level intelligence and potentially human-level AI, above the level of physical symbol systems. So in addition to applying knowledge to attain goals, I would enlarge the thought processes in the cognitive and rational bands to include all the abilities of human-level intelligence. For example, in addition to problem-solving, here are some thought processes which it seems clear involve thoughts being developed within the timescales of the cognitive band: Using and understanding natural language. As noted above, Newell gave an example (“please pass the salt”) of this task occurring in the cognitive band. Learning by creating explanations and predictions, using causal and purposive reasoning. Learning about new domains by developing analogies and metaphors with previously known domains. Imagining hypothetical situations. Spatial-temporal reasoning and visualization. Metacognition, analogical reasoning, causal and purposive reasoning. Thinking about the past, the present and potential futures. Reflecting about one’s thoughts. While it seems clear each of these happens in the cognitive band, one can envision variations which happen over much longer timescales, corresponding to the rational band. For example, reading and understanding a lengthy scientific paper may require one or more hours, involving many concerted actions of understanding natural language, as well as metacognition, etc. Writing a scientific paper may require several days of work, again involving many concerted actions of using and understanding natural language, metacognition, etc. While one may presume both these lengthy activities are motivated to achieve goals, and describe them as using knowledge to attain goals, such a description does not indicate the qualitative nature of thought processes involved, and loses information about the nature of mental processing happening in the rational band. 3. Why the rational and social bands are distinct On pages 152-153 of [8], Newell expresses doubts whether events that happen in social, historical, or evolutionary realms constitute higher bands of action. However, on pages 154-155 he suggests the social band can be considered distinct from the rational band and says social groups do not have the same degree of rationality as individual humans. He says the reason for this appears to be that individuals have more knowledge than they can rapidly communicate to each other, and that a social group cannot act like an individual with “a single body of knowledge” and “a single common basic goal structure.” I disagree with Newell’s reasoning, and will suggest different reasons for considering the social band to be distinct from the rational band. First, I’ll say why I disagree: The worldwide scientific community may be considered ‘superintelligent’ relative to any individual human: Individual scientists have developed ideas that are not understood by the average person, or even by scientists in other fields. This social group has developed and embodies more knowledge than any individual human, evidently. It seems clear that no modern scientist who has devoted the time needed to do research in his/her field of study would claim to understand all the ideas that have been developed in all other scientific fields and are being considered by the world scientific community. The world scientific community also shares a common goal, the expansion of scientific knowledge. Granted, individual scientists may think their fields are more important than others, so they may argue about prioritizing subgoals. Yet individual humans also have difficulty prioritizing their personal goals and an individual’s goals aren’t always coherent or consistent. Although the timeframe for natural language communication and disambiguation can be significant it does not prevent rational action by a group. Individual humans aren’t always rational, either.
4
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
713
Thus the reasons Newell gave for distinguishing the social band from the rational band are not valid. The existence and present state of the world scientific community contradicts Newell’s (1990) statements about limitations of social groups relative to individual humans. Social groups differ from individual humans in the following important ways Newell did not state, which distinguish the social band from the rational band of actions: Social groups can jointly know more than any individual. Because individuals have finite lifespans, presently limited to less than 12 decades, social groups can act over much longer timespans than individuals. Social groups can develop products and technologies which increase the range of physical actions and perceptions possible by individual humans, and by groups. Social groups can have goals that transcend an individual’s goals. Finally, social groups can create entities (such as money, corporations, laws, nations, etc.) that exist mainly via shared beliefs between humans. See [3] regarding the role of “inter-subjective” concepts in history. Considering all these points, the social band in many ways has greater import than the cognitive and rational bands. Humans are social animals, human intelligence is amplified by social interaction, and civilization is enabled by the social band. 4. The social band 4.1. Timescale and significance of social actions In pages 490-498, Newell [8] discussed the social band in more detail, as a topic “along the frontiers” for future research. He explained his timescale of days to months for actions in the social band by discussing “significant social interactions” among multiple humans, but noted that some significant interactions can last about an hour, such as conferences, games, meals, and lectures. He suggested ten minutes might be a lower bound for the timescale of social actions, by considering the activities of middle-level managers. It does not appear he defined the term ‘significant’. His comment that classroom sessions are not set at ten minutes indicates significance can involve whether something is accomplished by a social activity, if only to transmit knowledge via natural language. Yet nowadays someone can post a six minute video on YouTube with instructions about how to do something (e.g. how to replace strings on a guitar), and thousands of people can learn this from watching the video. The people who learn this may consider it significant, because it helps enable their future use of a guitar. Also nowadays, an individual can send a tweet which thousands or even millions of people read. It may take ten seconds to write the tweet, and one or two seconds to read it. Many people may consider the tweet significant. For example, some might buy or sell different stocks as a result. Arguably the significance of a social interaction is largely in the minds of those who participate in it: Two people establishing or maintaining a friendship can be significant, even if it involves only a few seconds of interaction. One person offending another can be significant even if it involves only a few seconds of interaction. One person giving another a telephone number or a stock tip can be significant, and only take a few seconds. Two people reaching a decision to do something together may be significant and can take only a few seconds. Human communication itself is a form of social action, and an event, even if no other physical action occurs directly as a result. So, in the following remarks about the social band I’ll use the terms ‘event’ and ‘action’ as synonyms.
714
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
5
4.2. Timescales of examples for social actions Using an open definition of ‘significance’, here are some examples I would suggest for social interactions along with typical time scales, with the disclaimer that these estimates are based only on informal observations and commonsense, i.e. they have not been validated experimentally: Newell’s example of one person asking another to pass the salt, and the other passing the salt, may be considered a social event on a time scale of 2 seconds. Other minimal, yet significant social interactions can happen in a few seconds, as noted in the previous section. Two people exchanging text messages – perhaps 1 to 2 minutes. Two people saying hello, perhaps shaking hands, chatting briefly, then saying “See you later” – 1 to 5 minutes. Two people chatting in a telephone call – typically 5 to 15 minutes, possibly an hour… Two or more people walking around a mall together, talking – typically 10 to 20 minutes. Two or more people at a table in a coffee shop, talking – typically 1 to 2 hours. A group of people having dinner together – typically 1 to 2 hours. A couple going out for a date – dinner and a movie – typically 3 to 4 hours. People attending a concert – typically 2 hours, possibly longer. People attending a dance – typically 3 to 4 hours. A ceremony for a wedding – perhaps 30 minutes. A ceremony for a funeral – perhaps 30 to 60 minutes. People attending a political rally, giving and hearing speeches – typically 3 to 4 hours. A congressional meeting – perhaps from 1 to 8 hours, with breaks. An election campaign – months to a year or two. A social fad spreading – weeks to months. A major change in social attitudes developing and becoming accepted – years to decades. From these examples it is natural to observe that although some events which happen at the social band can take place over longer time scales (days, weeks, months…), they are comprised of social events at much shorter time scales (seconds, minutes, and hours), and may involve cognitive and rational processing in individuals at time scales from seconds to minutes or hours. Much of what happens in the social band takes place in time scales of seconds, minutes, or hours. Anderson [1] discussed experimental evidence supporting a “Decomposition Thesis” that learning at the social band can be reduced to learning occurring at lower bands. He found evidence for the Decomposition Thesis down to the 10 millisecond level of the Biological Band. He also investigated other relationships between the social band and lower bands, relating to instructional outcomes and cognitive modeling, finding good evidence for relationships down to the unit-task level of the cognitive band. 4.3. Mental processes in the social band From the examples in section 4.2, it appears virtually everything that happens in the social band involves some degree of individual rational and cognitive processing, though much that happens also involves emotions and physical actions. It is clear from these examples that natural language generation and understanding are of major
6
C. /Jackson Science (2018) 710–716 P. C.Philip Jackson Procedia/ Procedia ComputerComputer Science 00 (2019)145 000–000
715
importance for many human social interactions: natural language is a bridge between individual minds in social groups. The social band is also vitally important for enabling individuals to learn a natural language. Children do not learn a natural language in a vacuum, they require interaction with adults and other children [9]. By enabling individuals to learn a shared natural language, the social band greatly augments an individual’s cognitive abilities: the individual gains a resource for acquiring knowledge from others, which is also a resource for individual thinking in the cognitive and rational bands. And it should be remembered that emotions are a language for the social band: emotions are the first language infants use for communication with parents. Emotions remain a language for communication throughout life, another way in which the social band rivals the importance of the cognitive and rational bands. 4.4. Modern technology and the social band There have always been phenomena in the social band which happen in timescales of seconds or minutes (people have always asked “Please pass the salt” in some language) but with the advent of modern technology the information bandwidth of the social band has increased exponentially. According to Wikipedia, it has only been about 100 years since loudspeakers were first available commercially for public address systems – before that, the human voice might only carry to an audience of perhaps a few thousand people in an auditorium.§ With loudspeakers, audiences of tens of thousands of people in an amphitheater could hear a single person speak – but this would only happen occasionally for a small number of speakers. With Twitter, thousands or millions of people can read a message from virtually any person, at virtually any time. Likewise, the advent of Wikipedia and YouTube exponentially increase the ability of individuals to share information publicly. Perhaps this increase in social information bandwidth may make the social band more comparable to the cognitive and rational bands. The social band approximates a ‘society of mind’ in the generalized sense (using a language of thought) described by Doyle [2], rather than the sense described by Minsky [6]. As discussed in section 3 above, the social band in many ways has greater import than the cognitive and rational bands: Humans are social animals, human intelligence is amplified by social interaction, and civilization is enabled by the social band. We may hope these exponential gains in bandwidth for the social band will yield amplifications of intelligence for civilization. 5. Implications for the Common Model of Cognition The above discussion has shown the social band is very important for human intelligence. This prompts the following suggestions for future research on the Common Model of Cognition: First, it could be interesting to perform experiments involving social groups of individual systems which each implement the Common Model of Cognition, supporting interaction between these systems using natural language (and in theory, also using emotions as a language). As a variation, it could be interesting to perform experiments in which one Common Model system represents an infant, learning a natural language by interacting with other Common Model systems that already know the natural language, representing parents or older children. As another variation, it could be interesting to perform experiments to see if Common Model systems could develop pidgin and creole languages, in situations such as those described by Pinker [9], pp. 32-39. Yet another variation would involve two Common Model systems that each understand different natural languages, learning to understand each other’s different languages.
§
For example, the Auditorium Theatre was designed to seat 4000 people in Chicago in 1889.
716
C. Jackson / Procedia Computer 145000–000 (2018) 710–716 P. Philip C. Jackson / Procedia Computer ScienceScience 00 (2019)
7
To model human-level cognition, it could be important to add a generalized society of mind (Doyle 1983) into the architecture of the Common Model of Cognition, using a natural language These suggestions are consistent with a separate paper [5] recommending use of a natural language of thought within a future version of the Common Model. 6. What is the timescale of inspirations? Perhaps some important cognitive and rational events can take much longer than seconds to minutes to hours. For example, August Kekulé once said he created his theory about the ring shape of benzene molecules after a daydream of a snake biting its tail, and that the vision followed years investigating carbon-carbon bonds. At other times Kekulé gave different anecdotes about conceiving the ring shape of benzene. [10] Yet the anecdotes suggest a question: Do inspirations that occur in a few seconds sometimes depend on subconscious processing over long periods of time? I don’t know if there is any quantitative evidence regarding this. Perhaps it can’t be studied in controlled psychological experiments, almost by definition. References [1] Anderson, John R. (2002) “Spanning seven orders of magnitude: a challenge for cognitive modeling.” Cognitive Science, 26: 85-112. [2] Doyle, Jon. (1983) “A society of mind – multiple perspectives, reasoned assumptions, and virtual copies.” IJCAI83: 309-314. [3] Harari, Yuval Noah. (2015) Sapiens: A Brief History of Humankind. HarperCollins Publishers. [4] Jackson, Philip C. (2018) The intelligence level and TalaMind. 2018 Conference on Advances in Cognitive Systems. Poster paper. [5] Jackson, Philip C. (2018) Natural language in the Common Model of Cognition. BICA 2018 Postproceedings: to appear. [6] Minsky, Marvin L. (1986) The Society of Mind. Simon and Schuster. [7] Newell, Allen. (1982) The knowledge level. Artificial Intelligence, 18: 87-127. [8] Newell, Allen. (1990) Unified Theories of Cognition. Harvard University Press. [9] Pinker, Steven. (1994) The Language Instinct – How the Mind Creates Language. HarperPerennial. [10] Wikipedia contributors. (2018) “August Kekulé.” Wikipedia, The Free Encyclopedia, 6 Jul. 2018.
ScienceDirect ScienceDirect
Procediaonline Computer 00 (2019) 000–000 Available at Science www.sciencedirect.com Procedia Computer Science 00 (2019) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
ScienceDirect Procedia Computer Science 145 (2018) 710–716
Postproceedings of the 9th Annual International Conference on Biologically Inspired Cognitive Postproceedings of the 9thBICA Annual International Conference Inspired Cognitive Architectures, 2018 (Ninth Annual MeetingonofBiologically the BICA Society) Architectures, BICA 2018 (Ninth Annual Meeting of the BICA Society)
Thoughts on bands of action Thoughts on bands of action Philip C. Jackson, Jr. * Philip C. Jackson, Jr. * 0F
0F
TalaMind LLC, PMB #363, 55 E. Long Lake Rd., Troy, MI 48085, U.S.A. TalaMind LLC, PMB #363, 55 E. Long Lake Rd., Troy, MI 48085, U.S.A.
Abstract Abstract This paper considers the ‘bands of action’ discussed by Newell in 1990. Reasons are given for a larger range of intelligent processing happen in cognitive and rational bandsbythan Newell Different reasons proposed This paper to considers thethe‘bands of action’ discussed Newell in described. 1990. Reasons are given forarea given larger from rangethose of intelligent by Newell for considering social band exist and be than distinct fromdescribed. the rationalDifferent band. Although suggested ~10proposed minutes processing to happen in thethe cognitive and to rational bands Newell reasons Newell are given from those might be a for minimal timescale for significant actions social from band,the reasons areband. givenAlthough for minimal timescales of seconds. The by Newell considering the social band to exist andinbethedistinct rational Newell suggested ~10 minutes importance of the social band in to create theincognitive and rational bands discussed. The impact of modern technology might be a minimal timescale forhelping significant actions the social band, reasons areisgiven for minimal timescales of seconds. The on the socialofband is described. Some potential therational Common Model of Cognition are suggested. A technology question is importance the social band in helping to createimplications the cognitivefor and bands is discussed. The impact of modern asked timescale of inspirations. on theabout socialthe band is described. Some potential implications for the Common Model of Cognition are suggested. A question is asked about the timescale of inspirations. © 2019 The Authors. Published by Elsevier B.V. © 2018 The Authors. by Elsevier B.V. This is an open accessPublished article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the the CC scientific committee of the 9th Annual International Conference on Biologically Inspired This is an open access article under BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Architectures. Architectures. Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Cognitive Architectures. Keywords: cognition, bands of action, rational band, social band, natural language, inspirations Keywords: cognition, bands of action, rational band, social band, natural language, inspirations
1. Newell’s ‘bands of action’ 1. Newell’s ‘bands of action’ In considering human cognition, Newell [8] discusses different time scales at which human action can occur. He In considering [8] discusses scales at whichbands human action can occur. and He groups these timehuman scalescognition, into the Newell following “bands ofdifferent action”,time saying different have phenomenal groups these time scales into the following “bands of action”, saying different bands have phenomenal and
* Corresponding author. address:author. [email protected] * E-mail Corresponding URL: http://www.talamind.com E-mail address: [email protected] URL: http://www.talamind.com 1877-0509 © 2019 The Authors. Published by Elsevier B.V. This is an open access under the CC by BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) 1877-0509 © 2019 The article Authors. Published Elsevier B.V. Peer-review under responsibility of the committee of the(https://creativecommons.org/licenses/by-nc-nd/4.0/) 9th Annual International Conference on Biologically Inspired Cognitive This is an open access article under thescientific CC BY-NC-ND license Architectures. Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Architectures.
1877-0509 © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 9th Annual International Conference on Biologically Inspired Cognitive Architectures. 10.1016/j.procs.2018.11.049
2
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
711
theoretical differences: Biological band Organelle (10-4 secs) Neuron (10-3 secs) Neural circuit (10-2 secs) Cognitive band Deliberate act (10-1 secs) Cognitive operations (1 sec) Unit task (10 sec) Rational band Tasks ranging from minutes to hours Social band Actions ranging from days to months Historical band† Actions ranging from years to millennia Evolutionary band Actions ranging from 104 to 106 years Except for the evolutionary band, these bands of action all correspond to real, physical events that happen due to processes within human brains. The evolutionary band involves physical events at the level of the human genome.‡ 2. The cognitive and rational bands After discussing operations by neural circuits, Newell [8] observed there is very little time available in the biological band to achieve cognitive behavior (p.129). To show this, he discussed the processing that must occur in a simple human interaction when one person asks “Please pass the salt” and another person responds in about a second. In the one second for a person to hear and understand "Please pass the salt" and then reach for the salt, there is only time to perform about 10 ‘deliberate acts’ of cognition. The one second includes speech recognition, natural language understanding, deciding whether to pass the salt, and then initiating an act to pass the salt (which could involve remembering where the salt was, looking for the salt, etc.). Newell noted it can also involve deciding whether the request to pass the salt was polite or impolite, and it might take longer to respond if the request was impolite. Newell observes that one consequence of the time scales required for biological processing is that the human computational system must be massively parallel to accomplish complex tasks in the timescale of the cognitive band. As an example of more complex cognitive processes requiring about 10 seconds for a unit task, Newell discusses playing ‘rapid-transit chess’, with rules allowing only 10 seconds per move. Newell (p.144) gives other examples of cognitive tasks, e.g. solving the CROSS+ROADS problem at about 2 seconds and solving a logic problem at about 8 seconds. Newell [8] writes (p.150) that as more time is available for solving problems, the system will have more time to find solutions using knowledge it already has. It may be described as a ‘knowledge-level system’, operating in the rational band. In the rational band an agent uses knowledge to achieve goals.
†
Figure 3-3 of [8] showed the first four bands. Figure 3-14 added the historical and evolutionary bands. Except where noted otherwise, page and figure numbers cited in this paper are relative to reference [8]. ‡ With current technologies, human actions may affect evolution in much smaller timescales.
712
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
3
In another paper [4], I give reasons why Newell’s definition of the knowledge level [7] is problematical. I discuss theoretical and practical reasons to recognize the existence of an ‘intelligence level’ corresponding to human-level intelligence and potentially human-level AI, above the level of physical symbol systems. So in addition to applying knowledge to attain goals, I would enlarge the thought processes in the cognitive and rational bands to include all the abilities of human-level intelligence. For example, in addition to problem-solving, here are some thought processes which it seems clear involve thoughts being developed within the timescales of the cognitive band: Using and understanding natural language. As noted above, Newell gave an example (“please pass the salt”) of this task occurring in the cognitive band. Learning by creating explanations and predictions, using causal and purposive reasoning. Learning about new domains by developing analogies and metaphors with previously known domains. Imagining hypothetical situations. Spatial-temporal reasoning and visualization. Metacognition, analogical reasoning, causal and purposive reasoning. Thinking about the past, the present and potential futures. Reflecting about one’s thoughts. While it seems clear each of these happens in the cognitive band, one can envision variations which happen over much longer timescales, corresponding to the rational band. For example, reading and understanding a lengthy scientific paper may require one or more hours, involving many concerted actions of understanding natural language, as well as metacognition, etc. Writing a scientific paper may require several days of work, again involving many concerted actions of using and understanding natural language, metacognition, etc. While one may presume both these lengthy activities are motivated to achieve goals, and describe them as using knowledge to attain goals, such a description does not indicate the qualitative nature of thought processes involved, and loses information about the nature of mental processing happening in the rational band. 3. Why the rational and social bands are distinct On pages 152-153 of [8], Newell expresses doubts whether events that happen in social, historical, or evolutionary realms constitute higher bands of action. However, on pages 154-155 he suggests the social band can be considered distinct from the rational band and says social groups do not have the same degree of rationality as individual humans. He says the reason for this appears to be that individuals have more knowledge than they can rapidly communicate to each other, and that a social group cannot act like an individual with “a single body of knowledge” and “a single common basic goal structure.” I disagree with Newell’s reasoning, and will suggest different reasons for considering the social band to be distinct from the rational band. First, I’ll say why I disagree: The worldwide scientific community may be considered ‘superintelligent’ relative to any individual human: Individual scientists have developed ideas that are not understood by the average person, or even by scientists in other fields. This social group has developed and embodies more knowledge than any individual human, evidently. It seems clear that no modern scientist who has devoted the time needed to do research in his/her field of study would claim to understand all the ideas that have been developed in all other scientific fields and are being considered by the world scientific community. The world scientific community also shares a common goal, the expansion of scientific knowledge. Granted, individual scientists may think their fields are more important than others, so they may argue about prioritizing subgoals. Yet individual humans also have difficulty prioritizing their personal goals and an individual’s goals aren’t always coherent or consistent. Although the timeframe for natural language communication and disambiguation can be significant it does not prevent rational action by a group. Individual humans aren’t always rational, either.
4
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
713
Thus the reasons Newell gave for distinguishing the social band from the rational band are not valid. The existence and present state of the world scientific community contradicts Newell’s (1990) statements about limitations of social groups relative to individual humans. Social groups differ from individual humans in the following important ways Newell did not state, which distinguish the social band from the rational band of actions: Social groups can jointly know more than any individual. Because individuals have finite lifespans, presently limited to less than 12 decades, social groups can act over much longer timespans than individuals. Social groups can develop products and technologies which increase the range of physical actions and perceptions possible by individual humans, and by groups. Social groups can have goals that transcend an individual’s goals. Finally, social groups can create entities (such as money, corporations, laws, nations, etc.) that exist mainly via shared beliefs between humans. See [3] regarding the role of “inter-subjective” concepts in history. Considering all these points, the social band in many ways has greater import than the cognitive and rational bands. Humans are social animals, human intelligence is amplified by social interaction, and civilization is enabled by the social band. 4. The social band 4.1. Timescale and significance of social actions In pages 490-498, Newell [8] discussed the social band in more detail, as a topic “along the frontiers” for future research. He explained his timescale of days to months for actions in the social band by discussing “significant social interactions” among multiple humans, but noted that some significant interactions can last about an hour, such as conferences, games, meals, and lectures. He suggested ten minutes might be a lower bound for the timescale of social actions, by considering the activities of middle-level managers. It does not appear he defined the term ‘significant’. His comment that classroom sessions are not set at ten minutes indicates significance can involve whether something is accomplished by a social activity, if only to transmit knowledge via natural language. Yet nowadays someone can post a six minute video on YouTube with instructions about how to do something (e.g. how to replace strings on a guitar), and thousands of people can learn this from watching the video. The people who learn this may consider it significant, because it helps enable their future use of a guitar. Also nowadays, an individual can send a tweet which thousands or even millions of people read. It may take ten seconds to write the tweet, and one or two seconds to read it. Many people may consider the tweet significant. For example, some might buy or sell different stocks as a result. Arguably the significance of a social interaction is largely in the minds of those who participate in it: Two people establishing or maintaining a friendship can be significant, even if it involves only a few seconds of interaction. One person offending another can be significant even if it involves only a few seconds of interaction. One person giving another a telephone number or a stock tip can be significant, and only take a few seconds. Two people reaching a decision to do something together may be significant and can take only a few seconds. Human communication itself is a form of social action, and an event, even if no other physical action occurs directly as a result. So, in the following remarks about the social band I’ll use the terms ‘event’ and ‘action’ as synonyms.
714
Philip C. Jackson / Procedia Computer Science 145 (2018) 710–716 P. C. Jackson / Procedia Computer Science 00 (2019) 000–000
5
4.2. Timescales of examples for social actions Using an open definition of ‘significance’, here are some examples I would suggest for social interactions along with typical time scales, with the disclaimer that these estimates are based only on informal observations and commonsense, i.e. they have not been validated experimentally: Newell’s example of one person asking another to pass the salt, and the other passing the salt, may be considered a social event on a time scale of 2 seconds. Other minimal, yet significant social interactions can happen in a few seconds, as noted in the previous section. Two people exchanging text messages – perhaps 1 to 2 minutes. Two people saying hello, perhaps shaking hands, chatting briefly, then saying “See you later” – 1 to 5 minutes. Two people chatting in a telephone call – typically 5 to 15 minutes, possibly an hour… Two or more people walking around a mall together, talking – typically 10 to 20 minutes. Two or more people at a table in a coffee shop, talking – typically 1 to 2 hours. A group of people having dinner together – typically 1 to 2 hours. A couple going out for a date – dinner and a movie – typically 3 to 4 hours. People attending a concert – typically 2 hours, possibly longer. People attending a dance – typically 3 to 4 hours. A ceremony for a wedding – perhaps 30 minutes. A ceremony for a funeral – perhaps 30 to 60 minutes. People attending a political rally, giving and hearing speeches – typically 3 to 4 hours. A congressional meeting – perhaps from 1 to 8 hours, with breaks. An election campaign – months to a year or two. A social fad spreading – weeks to months. A major change in social attitudes developing and becoming accepted – years to decades. From these examples it is natural to observe that although some events which happen at the social band can take place over longer time scales (days, weeks, months…), they are comprised of social events at much shorter time scales (seconds, minutes, and hours), and may involve cognitive and rational processing in individuals at time scales from seconds to minutes or hours. Much of what happens in the social band takes place in time scales of seconds, minutes, or hours. Anderson [1] discussed experimental evidence supporting a “Decomposition Thesis” that learning at the social band can be reduced to learning occurring at lower bands. He found evidence for the Decomposition Thesis down to the 10 millisecond level of the Biological Band. He also investigated other relationships between the social band and lower bands, relating to instructional outcomes and cognitive modeling, finding good evidence for relationships down to the unit-task level of the cognitive band. 4.3. Mental processes in the social band From the examples in section 4.2, it appears virtually everything that happens in the social band involves some degree of individual rational and cognitive processing, though much that happens also involves emotions and physical actions. It is clear from these examples that natural language generation and understanding are of major
6
C. /Jackson Science (2018) 710–716 P. C.Philip Jackson Procedia/ Procedia ComputerComputer Science 00 (2019)145 000–000
715
importance for many human social interactions: natural language is a bridge between individual minds in social groups. The social band is also vitally important for enabling individuals to learn a natural language. Children do not learn a natural language in a vacuum, they require interaction with adults and other children [9]. By enabling individuals to learn a shared natural language, the social band greatly augments an individual’s cognitive abilities: the individual gains a resource for acquiring knowledge from others, which is also a resource for individual thinking in the cognitive and rational bands. And it should be remembered that emotions are a language for the social band: emotions are the first language infants use for communication with parents. Emotions remain a language for communication throughout life, another way in which the social band rivals the importance of the cognitive and rational bands. 4.4. Modern technology and the social band There have always been phenomena in the social band which happen in timescales of seconds or minutes (people have always asked “Please pass the salt” in some language) but with the advent of modern technology the information bandwidth of the social band has increased exponentially. According to Wikipedia, it has only been about 100 years since loudspeakers were first available commercially for public address systems – before that, the human voice might only carry to an audience of perhaps a few thousand people in an auditorium.§ With loudspeakers, audiences of tens of thousands of people in an amphitheater could hear a single person speak – but this would only happen occasionally for a small number of speakers. With Twitter, thousands or millions of people can read a message from virtually any person, at virtually any time. Likewise, the advent of Wikipedia and YouTube exponentially increase the ability of individuals to share information publicly. Perhaps this increase in social information bandwidth may make the social band more comparable to the cognitive and rational bands. The social band approximates a ‘society of mind’ in the generalized sense (using a language of thought) described by Doyle [2], rather than the sense described by Minsky [6]. As discussed in section 3 above, the social band in many ways has greater import than the cognitive and rational bands: Humans are social animals, human intelligence is amplified by social interaction, and civilization is enabled by the social band. We may hope these exponential gains in bandwidth for the social band will yield amplifications of intelligence for civilization. 5. Implications for the Common Model of Cognition The above discussion has shown the social band is very important for human intelligence. This prompts the following suggestions for future research on the Common Model of Cognition: First, it could be interesting to perform experiments involving social groups of individual systems which each implement the Common Model of Cognition, supporting interaction between these systems using natural language (and in theory, also using emotions as a language). As a variation, it could be interesting to perform experiments in which one Common Model system represents an infant, learning a natural language by interacting with other Common Model systems that already know the natural language, representing parents or older children. As another variation, it could be interesting to perform experiments to see if Common Model systems could develop pidgin and creole languages, in situations such as those described by Pinker [9], pp. 32-39. Yet another variation would involve two Common Model systems that each understand different natural languages, learning to understand each other’s different languages.
§
For example, the Auditorium Theatre was designed to seat 4000 people in Chicago in 1889.
716
C. Jackson / Procedia Computer 145000–000 (2018) 710–716 P. Philip C. Jackson / Procedia Computer ScienceScience 00 (2019)
7
To model human-level cognition, it could be important to add a generalized society of mind (Doyle 1983) into the architecture of the Common Model of Cognition, using a natural language These suggestions are consistent with a separate paper [5] recommending use of a natural language of thought within a future version of the Common Model. 6. What is the timescale of inspirations? Perhaps some important cognitive and rational events can take much longer than seconds to minutes to hours. For example, August Kekulé once said he created his theory about the ring shape of benzene molecules after a daydream of a snake biting its tail, and that the vision followed years investigating carbon-carbon bonds. At other times Kekulé gave different anecdotes about conceiving the ring shape of benzene. [10] Yet the anecdotes suggest a question: Do inspirations that occur in a few seconds sometimes depend on subconscious processing over long periods of time? I don’t know if there is any quantitative evidence regarding this. Perhaps it can’t be studied in controlled psychological experiments, almost by definition. References [1] Anderson, John R. (2002) “Spanning seven orders of magnitude: a challenge for cognitive modeling.” Cognitive Science, 26: 85-112. [2] Doyle, Jon. (1983) “A society of mind – multiple perspectives, reasoned assumptions, and virtual copies.” IJCAI83: 309-314. [3] Harari, Yuval Noah. (2015) Sapiens: A Brief History of Humankind. HarperCollins Publishers. [4] Jackson, Philip C. (2018) The intelligence level and TalaMind. 2018 Conference on Advances in Cognitive Systems. Poster paper. [5] Jackson, Philip C. (2018) Natural language in the Common Model of Cognition. BICA 2018 Postproceedings: to appear. [6] Minsky, Marvin L. (1986) The Society of Mind. Simon and Schuster. [7] Newell, Allen. (1982) The knowledge level. Artificial Intelligence, 18: 87-127. [8] Newell, Allen. (1990) Unified Theories of Cognition. Harvard University Press. [9] Pinker, Steven. (1994) The Language Instinct – How the Mind Creates Language. HarperPerennial. [10] Wikipedia contributors. (2018) “August Kekulé.” Wikipedia, The Free Encyclopedia, 6 Jul. 2018.