An integrated theory of the mind. Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., Qin, Y. Psychological Review, 111(4). 1036-1060 (20041) : 19 2 23 Adaptive Control of Thought-Rational (ACT-R) ACT-R ACT-R Psychological review (Ungerleider & Miskin, 1982) (Squire, 1987) (Fodor, 1983) (Dehaene et al., 1999) (Warrington & Shallice, 1984) (Cosmides & Tooby, 2000) Newll (1990) 1
Soar...... ACT-R (Adaptive Control of Thought-Rational) 1970 ACT-R5.0 2007 6.0 Soar ACT-R ACT-R ACT-R - ACT-R 1 The ACT-R 5.0 Architecture (Figure 1) ACT-R5.0 visual module: manual module: declarative module: 2
goal module: 50ms (Braver et. al., 2001) (Nyberg et al., 1996) 5 3
ACT-R 2 The Perceptual-Motor System - ACT-R ACT-R - EPIC (Meyer & Kiera, 1997) - EPIC - ACT-R - : - color: red, vertical: top : (vertical: top, horizontal: middle) 4
: : : 185ms : (Schumacher, 1997; 2001) Visual-Manual Task [ ] [ ] [ ] Aural-Vocal Task 220Hz, 880Hz, 3520Hz 40msec 220Hz 1 880Hz 2 3520Hz 3 Visual-Mannual 279ms Visual-Mannual 445ms Visual-Mannual 283ms Visual-Mannual 456ms (Figure 2; Byrne & Anderson, 2001) VM +AV AV &VM AV AV (Figure 3) 5
3 The Goal Module (Anderson & Douglass, 2001) Disk 4 Peg C Disk 3 Peg B Disk 2 Peg C Disk 1 Peg B (Fincham et al., 2002) 16 fmri 4 (Figure 4) BOLD DLPFC: Partial region: premotar region: 4 The Declarative Memory Module 3+4=7 (Figure 5) 6
i (A i ): A i = B i + j W j S ji, (activation equation) B i : W j : S ji : j i W j n: W j =1/n S ji fan j : j S 2 S ji = S ln(fan j ) B i = ln( n: n j=1 d:.5 t j :j t d j ), (base-level learning equation) 1 P i = (probability of retrieval equation) 1+e (Ai γ)/s, s:.4 γ: T i = Fe Ai (latency of retrieval equation) F F 0.35e γ, A i = γ 0.35 (Pirolli & Anderson, 1985) 7
(hippie was in the park) 1-3 / 1-3 (Figure 6) 10 ACT-R (Figure 7) fan fan 5 Procedural Memory ACT-R Adaptive Control of Thought Control Adaptive ACT-R i (U i ) U i = P i G C i, P : C: G: i P i = eui/t n j euj/t t:.5 8
Sucesses: Failures: P = Successes Successes + Failures 1 Sucesses = θv + m, F ailures =(1 θ)v + n 1 θ θ (Friedman et al., 1964) 2 1; 90% 2: 10% 12 4 ACT-R 2 1 / 2 1 P 1 = 1 1+e (P2 P1)Gi/t P 1 : 1 P 1 : 2 1 (Figure 9) θ =0.5, V =2,G/t =2.25 P 1 = θv + m 1 V + m 1 + n 1 : 2 2 IF 9
THEN IF N THEN N : -7 : IF THEN 7 New Old1 0 θ =0 Δθ = a(p θ) P: Old1 a: New Old1 6 Putting It All Together: The Effects of Instruction andpracticeinadynamictask (Anderson, 2002) (Jones et al., 1999) postdict predict 10
6.1 The Dynamic Task: The Ani-Air Warfare Coordinator AAWC (Figure 10) : : : ALT: SPEED: : F6 [Track Manager] (Figure 11) Selection: Search: EWS (F10) Execute: Track-Manager(F6) 4 / 10 (F4/F7) Save (F1) 6.2 The Experiment 16 20 6 2 Apendix 40 (Fig. 8): 11
6.3 The Model 50ms EPIC 85ms d =0.5 s =.4 γ =1.0 F =1.0 t =.05 V =10 a =.05 200ms 6.4 The Speed Up in Task Performance Figure 12 12 13 Figure 13 : Figure 13 : (Figure 14) 12
6.5 Eye Movements (Figure 15) Selection Search Excute 6.6 Summary - 7 Putting It All Together: Tracking Multiple Buffers in an fmri data 7.1 The Experiment P 2 P 3 4 2 5 2 P 2 5 2 4 P 3 5 3 4 0-2 3-5-3-4 5 fmri 3 18 3 1.5s 13
BOLD 5 Figure 18 : BOLD : 5 BOLD Figure 18 : BOLD : Figure 18 : BOLD 0 : BOLD 7.2 The ACT-R Model 200ms 0 5 1 8 2 10 14
0 0 1 2 2 5 650ms 334ms BOLD t BOLD (Boyton et al., 1996; Cohen, 1997) B(t) =t a e t a 2 10 t=a BOLD x fmri BOLD ( ) t t x CB(t) =M i(x)b dx s M fmri s i(x) 1 0 0 Table 1 7.3 The Basal Ganglia BOLD (Figure 20) 15
8 General Discussion ACT-R ACT-R (Soar, EPIC) ACT-R ACT-R ACT-R ACT-R 16