Archive
Weekly QuEST Discussion Topics, 29 Jan
The weekly QuEST discussion is Friday, 29 January 2021, 1200 – 1300. We are using the GOOGLE HANGOUTS connection.
Information will also be available on the blog: https://qualellc.wordpress.com/2021/01/28/weekly-quest-discussion-topics-29-jan/
Meeting ID
Phone Numbers
(US)+1 917-310-4065
PIN: 644 051 134#
QuEST – 29 January 2021
QuEST 29 Jan 2021
Last week we discussed neurophysiological basis for breaking down the ‘semantic barrier’. Our Colleague Professor ‘Handsome’ George (used to be a professional wrestler) – suggested this is a key characteristic that will be required in 3rd wave AI. Such a breakthrough would enable real human-machine teaming.
To break the semantic barrier, multiple coherent representations might be needed to bridge between symbolic-statistical-human capabilities.
Our Colleague, Max, suggest if machines could acquire semantic systems in a similar fashion to humans then that would go a long way towards solving human-machine teaming issues. He also agrees that at the core of the semantic barrier problem is how multiple coherent representations bridge between symbolic-statistical-human capabilities. For example, linking the word form “bell” to the visual, auditory, somatosensory, and motor representations. Max proposes that the motor system (and in particular the efference copy signal) not only shapes perceptual representations but also effectively bridges the multiple coherent modality-preferential representations and provides a means for symbolic-referential linkages.
Cap
Weekly QuEST Discussion Topics, 22 Jan
Pulvermuller: How Neurons Make Meaning – Max K. Smith
For January 22nd, we will continue our larger discussion of how computational principles of representation in the human brain can inform development of true AI systems (S3). The paper that we will read to focus our conversation, “How neurons make meaning: brain mechanisms for embodied and abstract-symbolic semantics” by Friedman Pulvermuller seeks to address the semantic grounding problem. That is, “how words and other symbols are related to specific types of perceived objects and executable actions (Pulvermuller, 2013).” The general idea is that investigation of the simplest form of meaning making (grounding), which is assumed to be the linkage between perception and action, provides a foundation for the discovery of principles that underlie more complex forms of meaning making that are flexible, and begin to approximate “what it is like” to have/use knowledge. Pulvermuller proposes 4 types of semantic systems in the brain which all vary in degrees of overlapping and distinct purported mechanisms that instantiate them. Pulvermuller also makes an evolutionary/developmental argument about which semantic systems are primary or secondary and which semantic systems are likely to emerge from one another based on nueoimaging data.
Here are the proposed semantic systems (there may be more):
1. Referential semantic mechanisms – action-perception correlation and mutual inhibition
2. Combinatorial semantic mechanisms – Symbolic theft/secondary grounding
3. Emotional-affective semantic mechanisms – emotion (internal state)-action-symbol correlation through reinforcement learning and reward (external and internal)
4. Abstract semantic mechanisms – variable (low correlation) exemplars + out-of-sync delink in primary sensorimotor hubs + convergence hubs
To guide your reading there are a couple of questions that I would like you to consider:
1. At the outset, Pulvermuller proposes that, in the human brain, semantic processing involves multiple specialized brain areas rather than one semantic hub supported. Do you think that the decentralized organization of semantic systems reflects biological constraints, or a more fundamental principle of what is necessary to support meaning making systems (embodiment, or perception<->action cycles)?
2. Semantic or conceptual priming has been demonstrated to occur in the absence of attention and awareness. However, I don’t know if any examples of higher order semantic learning (i.e. learning that four-legged furry animal that barks is a dog) that is purely implicit or automatic. What may this tell us about the role of consciousness in semantic systems. That is, does there need to be a “what it is likeness” to be able to form a semantic system (link perceptual/emotional experience to symbolic representation)?
3. On page 466 there is an account of how a child may learn an abstract emotional word. Note that the learning is supported by a “language-teaching adult” (mature semantic system) which rewards the correct application of a word to the proper action schema. This account seems to suggest that for abstract semantics to be formed there needs to be an external shepherd of the learning process in the beginning. This may suggest that for any AI semantic system there needs to an established semantic system to guide semantic learning. Does this mean that humans will need to be “in the loop” during early learning? How do we reconcile the problem of humans accurately rewarding the right sign-meanings in an AI system when the action schema, which indicates a particular internal state, is very different from human action schema?
Additional Readings:
Pulvermüller, F. (2018). Neural reuse of action perception circuits for language, concepts and communication. Progress in neurobiology, 160, 1-44.
Cangelosi, A., Greco, A., & Harnad, S. (2002). Symbol grounding and the symbolic theft hypothesis. In Simulating the evolution of language (pp. 191-210). Springer, London.
Link to related paper – Grounding language acquisition by training semantic parsers
using captioned videos
Weekly QuEST Discussion Topics, 15 Jan
Kabrisky Lecture 2021
QuEST
14 Jan
Every January the QuEST group uses the first and second lecture of the calendar
year to present a ‘state of QuEST’ lecture in his honor as he was a founding
member of the QuEST group. This lecture is designed to bring anyone up to speed
on how we use terms and to communicate what we seek to accomplish.
QuEST is an innovative analytical and software development approach to
improve human-machine team decision quality over a wide range of
stimuli (handling unexpected queries and contextual adaptation) by
providing computer-based decision aids and decision engines controlling
agents that may be embedded in a platform interacting with the world
that are engineered to provide both intuitive reasoning and “conscious”
context sensitive thinking.
QuEST provides a mathematical framework to understand what can be
known by a group of people and their computer-based decision aids
about situations to facilitate prediction of when more people (different
training) or computer aids are necessary to make a particular decision.
Dr. Matthew Kabrisky was an Air Force pioneer and innovator. From Air Force
aviator in the 1950s to professor, mentor, and researcher, his discoveries paved
the way for many modern technological advancements. He developed theories of
how the human brain processes information to recognize visual objects. This work
directly led to the innovation of implanted electrodes for those afflicted with
diseases such as epilepsy and injuries that resulted in paralysis. He was the
leading international expert on the physiological symptoms of space adaptation
sickness, i.e., motion sickness. His research led NASA to a better understanding
and an approach to mitigate the effects of space environments on astronauts. His
research in the area of robust speech recognition laid critical foundations for
fostering the development of DoD and private industry products ranging from
voice activated controls in advanced tactical aircrafts, to aides for the disabled
and handicapped and industrial process control. In the 1990s, he helped lead a
team of engineers that developed the world’s most accurate breast cancer
detection system. This highly successful product has helped in the detection of
thousands of breast cancers before they would have otherwise been detected. Dr.
Kabrisky’s pioneering efforts paved the way for current innovations across the
Nation, the Air Force and at the Air Force Institute of Technology.
Weekly QuEST Discussion Topics, 8 Jan
Kabrisky Lecture 2021
Dr. Matthew Kabrisky was an Air Force pioneer and innovator. From Air Force aviator in the 1950s to professor, mentor, and researcher, his discoveries paved the way for many modern technological advancements. He developed theories of how the human brain processes information to recognize visual objects. This work directly led to the innovation of implanted electrodes for those afflicted with diseases such as epilepsy and injuries that resulted in paralysis. He was the leading international expert on the physiological symptoms of space adaptation sickness, i.e., motion sickness. His research led NASA to a better understanding and an approach to mitigate the effects of space environments on astronauts. His research in the area of robust speech recognition laid critical foundations for fostering the development of DoD and private industry products ranging from voice activated controls in advanced tactical aircrafts, to aides for the disabled and handicapped and industrial process control. In the 1990s, he helped lead a team of engineers that developed the world’s most accurate breast cancer detection system. This highly successful product has helped in the detection of thousands of breast cancers before they would have otherwise been detected. Dr. Kabrisky’s pioneering efforts paved the way for current innovations across the Nation, the Air Force and at the Air Force Institute of Technology.
Every January the QuEST group uses the first lecture of the calendar year to present a ‘state of QuEST’ lecture in his honor as he was a founding member of the QuEST group. This lecture sometimes takes more than one meeting as it is designed to bring anyone up to speed on how we use terms and to communicate what we seek to accomplish.
n QuEST is an innovative analytical and software development approach to improve human-machine team decision quality over a wide range of stimuli (handling unexpected queries and contextual adaptation) by providing computer-based decision aids and decision engines controlling agents that may be embedded in a platform interacting with the world that are engineered to provide both intuitive reasoning and “conscious” context sensitive thinking.
n QuEST provides a mathematical framework to understand what can be known by a group of people and their computer-based decision aids about situations to facilitate prediction of when more people (different training) or computer aids are necessary to make a particular decision.
Quest Group's Blog 