Core research question: Concurrency in human task processing - how do we think/process concurrently? Limitations of concurrency in humans? Computational model?
Self notes:
- Limits of multitasking depend on how many representationally-distinct pathways exist in the brain
- Recommended Structure: (target 10 pages)
- Abstract: A short overview of your entire paper. (0.25 pages)
- Introduction: Why is this research question important? (0.75 pages)
- Experiment design: How did you examine the research question? Describe the process of conducting the literature review/experiment/models/tools. For the computational model/tool, you should also describe in detail about your model/tool and how you implemented them. (1 pages)
- Results: What did you find through your study? (4 pages)
- Discussion: What are the implications of your results? (2 pages)
- Conclusion: Overview of results and discussion. (1 page)
- Limitation: What is the limitation of your study? What could future study explore? (1 page)
- My structure
- Introduction
- Design - historical, evolution, limits of concurrency, future (expansion)
- How does human brain process concurrent tasks?
- What is historical context of this topic?
- What are the limits of concurrency in human brains?
- Results
- History (1 page)
- Single channel theory, biases, various models
- Single processing unit
- Shared-memory vs message-passing
- Concurrency vs parallelism
- Stroop test
- Studies - list of studies done to find concurrency
- Perception and action as two fundamental types of cognitive tasks
- Cognitive function and motor function together, cognitive and cognitive together
- Models - connectionist
- Parallel information transmission
- Working memory model vs connectionist
- Differences from other mammals
- Limits of concurrency
- History (1 page)
- Discussion
- AGI
https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2015.01366/full
- Researchers debate whether multitasking involves serial or parallel processing
- Serial processing due to limited resources causes severe performance costs
- Central cognitive processing in Task 2 can occur parallel to Task 1, challenging single-channel theories
- Backward crosstalk logic and locus of slack logic are used to demonstrate parallel processing
- Studies show evidence of central Task 2 processing during Task 1 bottleneck stage, supporting capacity sharing assumption
- The concept of crosstalk challenges single-channel theories and highlights the interaction between central processes in multitasking.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4845905/ - Multiplexing, or the use of the same representations for different purposes, can introduce cross-talk among processing pathways
- The introduction of multiplexing decreases the number of tasks that can be executed at once
- The problem of mutual interference, or cross-talk, arises due to the multiplexing of representations; that is, the use of the same representations for different purposes — for example, the use of phonological representations for both encoding auditorily presented words, and for reading words out loud
https://journals.sagepub.com/doi/epdf/10.1080/00335557243000102
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2925295/ - Working memory is the small amount of information held in an easily retrievable state simultaneously
- Various models of working memory
https://www.sciencedirect.com/science/article/pii/S0732118X21000222
- research over the past 50 years has shown that serial-processing models are not an accurate account of human information processing
- Concurrency vs parallelism
- Shared-state model, message-passing
- Elementary information processes. Serial computing encouraged serial thought processing model
- We do have concurrency or parallelism otherwise motor functions would have interfered with higher-order cognition (thinking, decision making)
- Behavioural experiments can distinguish serial and concurrent processes
- Connectionist networks - parallel activation, parallel constraint satisfaction
- Shared state vs message passing - more message passing recently
https://www.cs.utexas.edu/~dana/Salvucci-PR08.pdf - Concurrency as threaded cognition
- Various combination of activities produce different results
- The first implication of our theory of threaded cognition is that concurrent multitasking does not require supervisory or executive processes to manage and schedule multiple task processes.
- Shared-resource model
Models of Central Capacity and Concurrency
- Single Channel Theory - old theory that says humans can process tasks in serial. Derived from then computers which were serial.
- Complex theories eveloped from von Neumann computer arch - several sensory inputs, one central processing operating one task at a time, outputs sent to peripherals, performed simultaneously.
- Concurrency in central mechanism:
- Kahneman's theory - a limit on central capacity - various ways to assign this limit concurrently (total capacity increases as demand of concurrent processes increases.)
- McLeod - fixed capacity - assigned as per the difficulty of the task
- Multiprocessing vs timesharing:
https://journals.sagepub.com/doi/epub/10.1111/j.1467-9280.1990.tb00067.x
- Behavioural experiments that show concurrency
https://www.nature.com/articles/s41467-023-43971-z
- Gap between parallel information transmission in humans and other mammals
https://en.wikipedia.org/wiki/Stroop_effect
- Stroop effect - used to test performance of
https://link.springer.com/article/10.1007/s00221-010-2429-6
- dual task effect (DTE)
http://www.isle.org/~langley/papers/icarus.csr17.pdf
- ICARUS cognitive architecture
- https://escholarship.org/uc/item/7542d3cm (concurrency in ICARUS)
https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2022.907121/full#B3
- Visual working memory concurrenytc
https://onlinelibrary.wiley.com/doi/pdf/10.1207/s15516709cog0000_12
- Concurrent Cognitive Task Modulates Coordination Dynamics
https://link.springer.com/article/10.1007/s10462-018-9646-y#Sec2
- Review of major cognitive architectures