Trying to implement mind on machines {machine consciousness, mind} {machine modeling of consciousness} {artificial consciousness} {synthetic consciousness} can help understand consciousness. People do not know what causes consciousness but can say something about machine ability to have experience. Perhaps, machines can develop, perceive, experience, have emotions, and have self [Brooks, 1997] [Brooks, 2002] [Brooks et al., 1998] [Churchland and Sejnowski, 1992] [Dreyfus, 1979] [Haugeland, 1985] [Kurzweil, 1999] [Lloyd, 1989] [Lloyd, 2003] [Lucas, 1961] [Simon, 1981] [von Neumann, 1958] [Weizenbaum, 1960] [Winograd and Flores, 1986] [Winston and Brown, 1979]. Machines can anticipate behavior, select actions, pay attention, make decisions, create art, simulate physical structures and events, learn facts and skills, simulate personality, plan actions, predict future situations, time events, use logic, and remember facts. Machines can do several things simultaneously. However, cognition is not consciousness.
possibility
Perhaps, consciousness cannot be in machines, because consciousness is brain illusions, must be forever mysterious, is too complex, is only concepts, or is incomplete and inconsistent.
analog
Perhaps, only analog machines can be conscious, because sensations appear to be analog.
level
Perhaps, machines can have only low-level subjective experiences.
biological tissue
Living systems have characteristics that differ from non-living physical systems [Schrödinger, 1944]. Perhaps, only biological tissue can give rise to mind and/or consciousness. Perhaps, consciousness requires structures or properties that only emerge in cellular electrochemical processes. Perhaps, consciousness requires structures or properties that only emerge over multiple reproductive cycles, because minds must have templates [Searle, 1992] [Searle, 1997].
complexity
Perhaps, to feel subjective experiences, machines must perform high-level functions, such as being kind, knowing beauty, being friendly, laughing, being moral, having goals, having motivation, being in love, using language, using metaphors, being creative, thinking about itself, solving new problem type, or having sensations. Perhaps, machines must perform complex tasks or carry out complex rules. Perhaps, machines need only to carry out many simple rules over long times to have intentions and goals [Nehaniv, 1998].
mind or soul
Perhaps, systems must create separate minds or souls to have subjective experiences. Perhaps, only minds can have or feel sense qualities. Such minds can differ from human minds, and sense qualities can qualitatively differ from human sense qualities. Besides creating mind, systems must create mechanisms for mind to affect physical world.
non-local interactions
Physical systems have only local immediate interactions among forces and other intensive quantities. Physical events at space and time points depend only on what is happening at that point. Living systems use information from different times and places to find patterns. Perhaps, consciousness requires non-local interactions.
robot
Perhaps, only robots can be conscious, because they must gather information, and machines without bodies cannot be conscious.
models
Machines that model self-consciousness can have parts that have continuous intelligent activity, such as solving problems, perceiving, and giving action orders. A part can register what is happening in other parts and analyze, remember, learn, and report, noting causes/inputs, results/outputs, efficiency, efficacy, and errors. Machine consciousness researchers include Axel Cleeremans and Douglas Lenat. Machines can have emotions or moods as reactions to system states, output, or input.
Machines can have complex world-and-self representations (Igor Aleksander) (Pentti Haikonen) (Owen Holland and Ron Goodman). Neural nets can recognize images and analyze them.
Machines can have complex world representations and imagine, plan, and predict (Igor Aleksander).
Machines can converse with humans, such as Intelligent Distribution Agent (Stan Franklin), which uses global workspace to integrate information and has lower-task modules.
Machines can have perceiving as reactive processes, thinking as analysis processes, and executive control for supervision or conflict resolution (Aaron Sloman and Ron Chrisley).
Virtual machines {CogAff schema} {H-CogAff architecture} can think about consciousness and combine cognition and emotion affect. Affect has three levels: reactive, deliberative, and management. Cognition has three levels: perception, analysis, and action. Therefore, affect and cognition can interact in nine ways.
Machines can have many simple components connected into modules and whole, as neurons connect, such as Corollary Discharge of Attention Movement (John Taylor) and Cyberchild (Rodney Cotterill).
ACT* model (John Anderson) has declarative knowledge in a semantic network with parallel spreading activation and has procedural knowledge in a production system with parallel matching.
Production systems (Hunt and Lansman) can have serial matching and semantic networks with parallel spreading activation.
Retrieving memories in parallel is unconscious, while using complex serial algorithms is conscious (Logan, Stanley, and Neal) (Hesketh).
Implicit connectionist relations and categories are like unconsciousness, and explicit associations and spatiotemporal context are like consciousness (Bower).
CLARION model (Sun) uses distributed representation, for not directly accessible information, and symbolic local representation, for directly accessible information.
Systems can require representation activation above threshold (Bowers).
Systems can use production rules or perception symbols as representations and work with several "chunks" or combine simple chunks into new complex chunks (Servan-Schreiber, Anderson, and Rosenbloom).
Systems can be dynamic and go through transient states or reach stable-state attractors, in which everything is consistent (Mathis and Mozer) (O'Brien and Opie).
Outline of Knowledge Database Home Page
Description of Outline of Knowledge Database
Date Modified: 2022.0224