1. 마스터 제목 스타일 편집 마스터 텍스트 스타일을 편집합니다 둘째 수준 셋째 수준 넷째 수준 다섯째 수준 Queueing Network Model of Elementary Mental Processes Yili Liu 이 석 원이 석 원

2. Division of Information Management Engineering UI Lab. 2 Abstract Examines the use of reaction time (RT) to infer the possible configurations of mental system and presents a class of queueing network models of elementary mental processes. RT model - temporal issue (discrete vs. continuous information transmission) - architectural issue (serial vs. network arrangement of mental process)

3. Division of Information Management Engineering UI Lab. 3 Introduction Why is there delay between stimulus presentation and response initiation? Model for RT - temporal dimension (discrete vs. continuous information transmission) - architectural dimension (serial vs. network arrangement of mental process) Grain size

4. Division of Information Management Engineering UI Lab. 4 Classification of RT model

5. Division of Information Management Engineering UI Lab. 5 Literature Review for RT model 1.Discrete transmission and Serial arrangement - The Subtraction Method (Donders) - The Additive Factor Analysis (Sternberg) - The General-Gamma Model (McGrill & Gibbon) 2.Continuous transmission and Serial arrangement - The Cascade Model (McClelland) - The Queue-Series Model (Miller) 3.Discrete transmission and Network arrangement - Intercompletion Times (ICTs) (Townsend) - PERT Network (Schweikert)

6. Division of Information Management Engineering UI Lab. 6 4. Continuous transmission and Network arrangement - Queueing Network Model (Liu) Queueing Network Models for RT in General Form Series Queues Fork-Join Network Simon-Foley Network Closed Queuing Network

7. Division of Information Management Engineering UI Lab. 7 General Description of Queueing Network Model 1. General Assumption and Notions - 8 assumptions - 2 notation sets 2. Stimulus Components as Customers 3. Component Arrivals, Services, and Routing Characteristic

8. Division of Information Management Engineering UI Lab. 8 4. Product-Form Networks (Jackson Networks) - Poisson Arrival / Exponential Service / State-Independent Routing General Description of Queueing Network Model

9. Division of Information Management Engineering UI Lab. 9 5. Sojourn Time in Queueing Networks - sojourn time : total time for a customer to traverse - overtake-free path in a product-form network General Description of Queueing Network Model

10. Division of Information Management Engineering UI Lab. 10 - not overtake-free path in a product-form network General Description of Queueing Network Model

11. Division of Information Management Engineering UI Lab. 11 6. RT as Network Sojourn Time - 가정 : N 개의 stimulus component 중에서 M 개가 response unit 에 들어오면 response 가 나타난다. - total RT - input node in a discrete and continuous network General Description of Queueing Network Model

12. Division of Information Management Engineering UI Lab. 12 Series Queues as a Model for RT 1. Network Sojourn Time in Series Queues - series queue : node 들이 1 열로 배열하고, 진행은 한쪽 방향으로 만 가능하다.

13. Division of Information Management Engineering UI Lab. 13 - Continuous-series Queueing Network - 만약 K 개의 node 의 mean service rate 가 다른 모든 것과 서로 다를 경우 감마분포형태로 표현가능하다 - Continuous-series Queueing Network 에서 RT 는 감마분포를 한다 Series Queues as a Model for RT

14. Division of Information Management Engineering UI Lab. 14 2. Serial Discrete-Stage Model of McGrill and Gibbon - transmitted as an indivisible unit from the first node to the K-th node - Serial Discrete-Stage Model 에서 RT 는 감마분포를 한 다. Series Queues as a Model for RT

15. Division of Information Management Engineering UI Lab. 15 3. Additive Stages Model of Ashby and Townsend - 가정 : RT process can be decomposed into a number of additive stage. - RT can be decomposed into a number of ICTs. - 하지만 K 번째 ICT 는 알 수 없다. 그래서 K 번째 ICT 를 지수분포를 한 다고 가정하면 RT 는 … Series Queues as a Model for RT

16. Division of Information Management Engineering UI Lab. 16 - If there are 2 levels of 2 experimental factors, A and B - selective influence Series Queues as a Model for RT

17. Division of Information Management Engineering UI Lab. 17 4. Serial Continuous Model 1) McClelland’s Cascade Model - Cascade Equation - all of the units at the same processing level have the same activation function Series Queues as a Model for RT

18. Division of Information Management Engineering UI Lab. 18 - effect of experimental manipulation : rate constant, asymptotic level 2) Miller’s Queue Series Model - 가정 : the stimulus is regarded as consisting of a number, M, of distinct components, which are serviced by a series of processing nodes. - evaluated through a novel application of PERT method Series Queues as a Model for RT

19. Division of Information Management Engineering UI Lab. 19 Mean RT in Fork-Join and Feedback Queueing Networks Series Queueing Networks vs. Non-Series Queueing Networks 1. Fork-Join Queueing Networks - when the service requirements of a customer do not have to be processed in strict sequence. - fork node and join node

20. Division of Information Management Engineering UI Lab. 20 - network sojourn time - when associated variable Mean RT in Fork-Join and Feedback Queueing Networks

21. Division of Information Management Engineering UI Lab. 21 - expected values has an upper bound - expected value Mean RT in Fork-Join and Feedback Queueing Networks

22. Division of Information Management Engineering UI Lab. 22 2. A Single-Server Feedback Queueing System - feedback system is not overtake-free and the order of customer arrival is not preserved in the order of their departure form the system - sojourn time Mean RT in Fork-Join and Feedback Queueing Networks

23. Division of Information Management Engineering UI Lab. 23 - feedback system with N identical server - feedback system in discrete vs. continuous system Mean RT in Fork-Join and Feedback Queueing Networks

24. Division of Information Management Engineering UI Lab. 24 Conclusion The Queueing Network Theory has the capacity to model a greater variety of processing system