Contract Monitor. MILTON E. WOOD ..... Hierarchical Clusters - Split-Plane Concepts - IPs..1q. 3 ... Hierarchical Clusters m Strafe Concepts - IPs ....... 21. 5.
AFHRL.TP-81-46
AIR FORCE I__ '"
C)
-
'
HFOR
U M
A
STRUCTURES OF MEMORY CRITICAL FLIGHT INFORMATION
By Roger W. Schvaneveldt Timothy Durso FrancisE.T.Goldsmith Kenneth Maxwell
Hector M. Acosta Department of Psychology University State MexicoNew ~Las Crues, Mexico 8W03 New
ymm•
Richard G. Tucker d9R!C479th
TlW/Training Analysis Holloman AFB, New Mexico 88330
E•
E
Williams Air Force Base, Arizona 85224
S
June 1982
U R 0
R
E S
OPERATIONS TRAINING DIVISION
Interim 1 May 1980 - 31 July 1981
Approved for public release, distribution unlimited.
LABORATORY AIR FORCE SYSTEMS COMMAND
S...
.K->
A
BROOKS AIR FORCE BASETEXAS 78235
NOTICE Wle', (Governrliti drawitagm, "peCifieationb, or other data are used for any purpose oilehr than iii(vtnnection with a definitely Governinenlt-related procurement, tile United State, Government ineurs no responsibility or any obligation whatsoever. The fact that the Government may have formulated or in any way supplied the said drawings, specificat ionh, or other data, is not to be regarded manner construed, as licensing the holder, or any by implication, or otherwise in any other person or corporation; or aa .onveying any rights or permisnmion to manufacture, use, or ,ell any patented invention that may in any way be related thereto. The Public Affairs Office has reviewed this paper. and it is releasable to the National 'rechnical Information Service, where it will he available to the general public, including foreign nationals, This paper has been reviewed and is approved for publication. JOSEPH C. I)EMAIO Contract Monitor
MILTON E. WOOD, Operations Training Teelhnical Division Director RICHARD G. CRONQUIST, Colonel, USAF Chief. Operations Training Division
p.J
unclass'ifiedl SECURITY CLASSIFICATION OF THIS PACE (When, Data, Entered),__________________
REPOT PT
GOVT ACCESSION NO.
USmER"2
3. RECIPIENT'S CATALOG NUMBER
S. TYPE OF REPORT & PERIOD COVERED
4.TITLE (and Subtitle)
$II'TI'RES OF MEMORY. FOR CRITICAL FLIGHT INFORMATION
I May I1980 -. 31 July I'98l 6. PERFORMING 01G. REPORT NUMBER S. CONTRACT OR GRANT NUMBER(&)
7. AUTHOR(s)
F3 -0CHI 3058-(H4
KennethI Maxwell Iliclor M. Acoista Richard G. Tucker
Roger W. Selhvaneveldt Tinloily E. (;oldsnlitil Francis T. D~urso 9.
READ INSTRUCTIONS COMPLETING. FORM
AGEBEFORE iMUMETA.ION
IDepartmnent of Psychology6102 New Mexico Stale University Las Cruces. New Mexico 884N)3 It.
PROGRAM ELEMENT. PROJECT, TASK
10.
PERFORMING ORGANIZATION NAME AND ADDRESS
AREA &WORK
612 .ZII 12.
CONTROLLING OFFICE NAME AND ADDRESS
Brooks Air Force Base. Texas 78235 14. MONITORI NG AGENCY NAME &AODRESS(if
Ajilkulls1
Trinn . III %ir
1`4erv
13.
NUMBER OF PAGES
'118 different from Controlling Office)
IS.
iIII Ba~se.
REPORT DATE
June 1982.
HIQ Air Force Human Resources Laboralorv (AFSC)
IOperalin
UNIT NUMBERS
SECURITY CLASS. (of thie report)
U!nclassified
Arzn
8""
IS..
DECLASSIFICATION/ DOWNGRADING SCHEDULE
1.DISTRIBUTION STATEMENT (of this Report)
Apprvedforpublic release: distribution unlimited.
17.
DISTRIBUTION STATEMENT (of the abstract enteted'in Block 20, If different fromn Report)
IS.
SUPPLEMENTARY NOTES
I.KEY WORDS (Continu, on reverse side it necessary and Identify by biock number)
critical flighlt infoirmation devl'eopmentfl of flight concepts fly.ing training stii-neof pilots' concep1.jts ABSTRACT "'~20.
(Continue on reverse sids It necessary snd~ fdentify by block number)
SThis paper reviews work that has been done on defining 'andl measuring conceptual structures of critical flight iiiforniation in Air Force fighter pilots. Individuals with widelwy vIarying fliglht experiencee were tested. Cognitive structures were defined byvanalytic procedures: i~g.. Multidimensional Scaling (MDS) and General Weighted Networks (WWN). The. NPS' analy~sis showed that the level of flying expericnce can be predicted from tlhe p~ilot~s I'olcetua srucure :re WN analysis led to tihe identification of specific points of agreement and disagreement in thel colleeptual organization of novice andl expert pilots. Pilots do have measurable cognitive stru~cture.-; for organizing flight-related informnation. Phese struc~ures are mecasurably different for individuals with dlifferent flight experience. The tlrcllliques employedl iti 1114'research produce (lescrijptions of conceptual structure thlat may have appli('ations in
DD i¶jAN 73 1473
ORMt
EDITION OF I NOV 65 15 OBSOLETE
Unc~lassified SECURITY CLASSIFICATION OF TH.IS PAGE ("ien Data Entered)
individual difftr'n4 tile mirning program of fightrr PIE0I8 Aiid in assouminig
1114Ii-w~4vf'opmrlIen of 1)1)Voprialf
C¶onceptual strulctures4,
II(!ulmi
fiied
SIUCJIRITY CLASSIFICATION COr Tu-4- PAOF(hurbej
1;t
liap
Jnne 110112
AFHRL Technical Paper 81-46
STRUCTIIIIIES OF MEMORiY
FOI CRITICAL FilI
;IT
INIFOIRMATION
Bly Roger W. Sehvaneveldt Timothy E. Goldsmith Francis T. Dursn Kenneth Maxwell Hector M, Aeonta Department uf Psiyeholhgy
New Mexico Stasto Univwroity Las Crues. New MeWiio 880013
Richard G. Tucker v9th TTr l'arining Aktaival Ho+ lllnistn AFU, New Marse,, h6330)
•i
.,,col
i~
Iti
Reviewed by Thomas M. Longridge
Chief, T'raining Effectiveness Sectionn
,•
,Operations Training Division Williams Air Fnrce Base, Arizona 85224
S~Submitted ..
for Publication by 1I
Thomitas It. (pray
SChief,
Training Technology Branch Operations Training Division Williams Air Force Base, Arloona 85224
This paper Impublished in the interest of scientific and technical infornmatiuon exeliainget the bine reprarelh reported represents an early stage In the development of eognitive viructures fur flight-rclated inforoelihn,
I
i • '+-•
. .
..
.o
... ,
•
,•
+
:+i
-•11... ,,
,
SUMMARY Objective The ol~jOliv'4 WNK intformattiotn concerning [lie organivtation of pilots' inetnories for eritival flight relatled it oingilon mla vstrip io oflly systenia Iic diffe rc ires; ill ineillory st ructuren that were related to differelgwes ill
Bockground/Rativenale Heslearch Into natural language lita suggested that retrieval of information fromt memory is affected by tihe tirganisaitiun of tmemnory. More rapid and effective retrieval can result from a inore efficient and economtical storage. st1ructure, The organisation of Intformationun memory cah)ove a critical impact on flying performtance. Uinderstanditng how critical information is organised in memory can be extremely useful to training program dosignrs antd evaluators asl well as Instructors and others interested in increasing lthe effectiveness of thia pilot. uIreraft stysemi. Knowledge of how Individuals develop systems for organlaing critical informatiotn can be used to tailor trtaitnittg wystent to provide students the conceptual framework that will load to optimal learning. It may also p~rovidei a useful selection toot by allowing Instructors to deternmine which Itndividuals have mnastered lte1 prerequisitie rottctipts for muccessi in a particu lar training program, Approach Two Nets of stimtulus tnstwrials displayed ott thea console of a Terak 8510/A inticroconttputer were premented to four groups of officersi A~r National Guard pilots (CPa), Fighter Lead-in Instructor Oilots (11's). recent U1ndergraduate Pilot Trie.ning (UPT) graduates. and Instructor Weapons System Officers (WSOa). Three statistical titelniqueN (htierarcihical cluster analysis, ,ulti-dimettaional scaling (MDS) and general weighted network were timed toantalyze tlia data and describe lthe cognitive structure of thea groups studied. specific$ Subjeets were nine A-7 ONs fron tlite Colorado Air National Guard, seven Fighter Lead-in INs, four Fighter Lin'ad-itt Inotruclor WS0s, and 17 recent UPT Graduates. The CGu each liad In excess of 2000 hours flying tintte. The IN liad 1200)-49(X) hours, A~ ilta UPT graduates had about 200 hours. Although thea GPs and IN liad that tlte GPs litad little Instructor experience roughtly sittllar tniounts of flying timie, their experietnce differed in% whilrteilaINs liad relatively less operational experience, The WS0s hatl 800-30(M) houro. Two concepitual sets were, investigated. One, the low-angle istrafe, was related to only a single maneuver. The were gene~nted by thte other Not dealt Witlti a class of malneuversi tilea split-plane miatteuvers. The stimuitus seels 4.lwitierlttewrN working Withi senior INs at Hollomnan APB, Populations of flight related terms were assetnbled through interview" Witlti INs at Ilolloman AFB, These were condensed to one set of 30 stimulus items for split-platte sod :4(1 for st~rafe. -
*
~Sulijeerts performed a self-pared rating task in which all pair-wise combinations4 of lthe itenms int eacht set were rated for Fsitnilarity. Similarity ratings were treated as distiences in conceptual spate. lit addition, ltme 1.11s gave a ruling of thteir famtiliarity Wilittlie tertms, This was done to ensure that thea UPT graduates haed at least a mininintal degroe of famtiliarity with thea terms. Aitalysesi showed tlite tnore experienced pilots to have conceptual structures that were better developed, mnore swophlisticated, and more oconontical tItaon the UPT graduates. The WS(6 were also found to htave conceptual structures which differed fromt tltose of the piolos. A pattertn recognition algorilhnt was applied to lthe MIJS solutions attd to ilie raw ratitng data to see if thet gmtmpsA could his distinguished ol thea basis of their conceptual structures. This program searched for a pattern or
proltotlype which ltaracleried the conceptual structure of members of a group. The groups were found to have dlfferenn coneoplual structures, with the mare experienced individuals ahowing a mlightly greater tendency to cluster about tho group prototype, while lite recent UPT graduate* tended to be more diverse.
i
Conchaelusion/Recommendatons I, Pilots do have measurable cognitive structures for remembering and recalling flight related information. 2.
Cognitive• otruclurem show meamurable differencew an a funrtion of flyinig oxperience. Experienced pilot,
e*hihlo miore efficient and economical organization of flight related Informnation than do lei experienced pilots. The WSO alhowed a memory structure that differed from thtat of Lhe pilots. 3.
'rT.happroalh
iidividuall'
used in the present research con provide a useful tool for lookIng at differenoos in undetanding of flying tuaksi, Buh a tool may have application both %Pv aaesiving inoiSvlduol
dIfereoncel In the development of conceptual understanding during learning and for l•okiall' ah,. -fetWivenes. of training programs In conveying critical flight related concepts to students.
Alli
I,•
prototype which characterized the conceptual structure of members of a group. The groups were found to have different conceplual struciures, with the more exp.ienceid individuals showing a slightly greater tendency to cluster about the group protolype, while tile rcuent UPT graduates tended to be more diverse.
Coueltisions/Recozmmendations I. Pilots do have ineasurable cognitive structures for remembering and recalling flight related information. 2. .Cognitive structures show measurable differences as a function of flying experience. Experienced pilots exhibit imore efficient and economical organization of flight related information than do less experienced pilots. The WSOs showed a memory structure that differed from that of the pilots. 3. The approach used in the present research can provide a useful iool for looking at differences in iindividuals' understanding' of flying tasks. Such a tool may have application both for assessing individual differences in the development of conceptual understanding during learning and for looking at the effectiveness of training programs in conveying critical flight related concepts to students.
Li_
I
TABLE OF CONTENTS
INTRODUCTION
.
. .....
............
.....................
........
5
GENERATION OF STIMULUS MATERIALS ...............................
SIMILARITY-RELATEDNESS RATINGS ...............................
18
. ...........
HIERARCHICAL CLUSTERING .......................
.
MULTIDIMENSIONAL SCALING ..................................
PATTERN RECOGNITION ANALYSIS OF CONCEPTUAL STRUCTURES ......... .............
GENERAL WEIGHTED NETWORKS ........
.....
314 58
GENERAL DISCUSSION.*.........................................84 REFERENCES ............. APPENDIX ......................
0
................................ *
e . . .s. .m
. . ... * .s t.
"".
. . .p.e.s.s
PA
. 93
LIST OF ILLUSTRATIONS Figure
Description
Page
I
Hierarchy of Natural Language Concepts ...............
2 3 4 5
Hierarchical Hierarchical Hierarchical Hierarchical
6 7 8 9
MDS Solution MDS Solution MDS Solution MDS Solution -
2
Clusters - Split-Plane Concepts - IPs..1q Clusters - Split-Plane Concepts - UPs..20 Clusters m Strafe Concepts - IPs ....... 21 Clusters - Strafe Concepts - UPs ....... 22 Split-Plane Concepts - IPs ........... Split-Plane Concepts - UPS ........... Strafe Concepts .... Strafe Concepts - UPs...
25 26 27
10
Minimum Connected Network
11
Minimum Elaborated Network
12
Cycles in the Network - Split-Plane - IPs ...........
13 14 15
Minimum Elaborated Network - Split-Plane - UPs ...... 63 Minimum Elaborated Network - Strafe ,.IPs.......64 Minimum Elaborated Network - Strafe UPs ........... 65
16
Assemblies in the Network - Split-Plane - Irs.......66
17
Clusters in a Multidimensional Structure........87
Split-Plane - IPs .......
-
-v[-
-
60
Split-Plane - IPso.....61 62
LIST OF TABLES Table
Description
Page
I 2 3
Concepts from Split-Plane Maneuvers .................... Concepts from Low Angle Strafe Maneuver ............... Essential Aircraft Terms ...............................
7 8 9
14 5
Subject Groups for the Rating Task ................ Flying rime for Individual Subjects_................
11 1;'
6 7
Familiarity of UPs with Each Concept ................ Familiarity of Individual UPs with all Concepts .......
14 15
8
Rating Score Correlations for, IPs ...................
9
Concept Clusters for IPs and UPs - Split-Planeo....... 23
10
Dimensions in the Multidimensional Solutionsu.........30
11 12
Within and Between Group Correlations - Split-Plnne...32 Within and Between Group Correlations - Strafe ........ 33
13
Classifications with the Minimum-Distance Classifier..38
14
Distances between Group Prototypes....................140
15 16
Analysis of Individual IPs - Split-Plane..............1 2 Analysis of Individual OPs - Split-Plane.............. 1 43
17
Analysis of Individual IWs - Split-Plane ...........
18 19 20 21 22 23
Analysis Analysin Analysis Analysis Analysis Analysis
24 25 26
Predictions of Group Membership - Split-Plane.........'3 65 Predictions of Group Membership - Split-Plane ........ Prediations of Group Membership - Strafe .............. 55
27
Pairs of Concepts that Discriminate IPs and UPs .......
28
Dominating Concepts andAssemblies..................71
29
Comparison of Novice and Expert Concept
30
Erroneous Links I.n the Novice Network............
of of of of of of
Individual Individual Individual Individual Individual Individual
UPs UPs IPs lWs UPs UPs
-vi.i-
44
- Split-Plane - Ratings.,.. 115 - Split-Plane - MDS ...... 116 -strafe ................. '47 4 -Strafe................. 8 - Strafe - Rntings ......... 49 - Strafe - MDS ............ 50
.....t s.......
57
r INTRODUCTION In the past decade, experimental psychologists have generated a considerable body of theory and data concerning the organization and retrieval of knowledge in human memory. This research area (which has come to be known as semantic memory) has concentrated largely on the study of natural categories and their members (e.g., birds, mineral,, geological formations). One of the first theoretical proposals in the area was suggested by Collins and Quillian (1969) following the lead developed by Quillian (1969) in the form of an intelligent, questionanswering computer system. Two important structural principles were ombodied in the theoretical analysis offered by Collins and Quillian: hierarchical orRanization and cognitive economy. The hierarchical principle refers to the proposal that concepts are stored in memory as nodes in a network with each node having labeled links to other nodes that repre.ent superordinate concepts. For example, the node representing the concept "robin" would have a particular kind of link (ISA) to the node representing the concept "bird." The hierarchical scheme requires that each concept only be connected to its immediate superordinate and not to more general concepts (e.g., robin is directly connected to bird but not to animal). The hierarchy also provides a basis for inferences about facts not learned directly. If the structure contains the facts that "an A is a B" and "a B is a C", then it can be inferred that "an A is a C." The principle of cognitive economy refers to the way in which properties of concepts are represented in the memory system, In particular, properties are stored at the highest possible level of the hierarchy. This means that properties pertaining to all members ot' a particular category need to be stored only once with a link to the node representing the category. For example, the property "ha:i wings" would be stored with the concept bird rather than with each particular type of bird. Thus there is an economy of storage. Both principles are illustrated in the network structure shown in Figure 1. Collins and Quillian (1969, 1970) presented evidence favoring the hierarchical theory of human memory structure, They showed that people were faster in verifying true sentences relating concepts near in the hierarchy (e.g., A robin is a bird.) compared to sentences relating more distant concepts (e.g., A robin is an animal.). Simi tar results were found for' sentences asserting property relations. Those sentences were also verified more slowly when the noun and the property were further apart in the hierarchy.
i
0-
uI)
o
I
mj z 0 a >
0 ucj
-
j -D
-j
:U< L wj ui
V
C) 0
xI
0
=U
uj
L
oz0C
0 0
0cccc Muui R Luw U-
cc
cr
wU
LU
2L
DJ
a:J
-J 0
Z
CL LL 0
Z
C 6 >- 0 - V% L, -
LL
UJ J JC LU z
>1
0
e
P L)
(
Lu
X z V)
ui
>-
-JV
-
-
0
I-
LU
-J
0L0cI-j Z
0w
5 LQIC L C-
L uj
r. -j6L
C Vuu
wLU V
:t CC3u 0(D = ci:
-20-
.2 4)jL
t
z
Uu 0.
r
0
La
2j
1u0
cc
z
0
U. 8t0a0c§
-2D-
NZ
z-
z~~ ui D ~~L
RE
z
w
c
I.-
-
-
