Human Processes in Intelligence Analysis: Phase 1 Overview

8/8/2019 Human Processes in Intelligence Analysis: Phase 1 Overview

1/74

RESEARCH REPORT 1237

HUMAN PROCESSES IN JUL T8INTELLIGENCE 9PHASE I OVERVIEW"Robert V. Katter2Christine A. NatgonmeryJohn R. ThompsonOperatIrks Systew, Inc.

HUMAN FACTORS TECHNICAL AREA, ARIOFFICE of DEPUTY CHIEF of STAFF-SYSTEMS, INSCOM

ARMY RESEARCH INSTITUTE US ARMY INTELLIGENCES-'rtheEHAVIORAL and SOCIAL SCiU4CES and SCRT OMNLu 0 1 Eisenhower Avenue Arlington Hall Station

xariVirginia 22333 Arlington, Virginia 22312

S~80 7 10 02 2jk.. d


2/74

U, S. ARMY RESEARCH INSTITUTEFOR THE BEHAVIORAL AND SOCIAL SCIENCESA Field Operating Agercy under the Jurisdiction of theDeputy Chief of St3ff for Personnel

FRANKLIN A. HARTJOSEPH ZEIDNER Colonel, US ArmyTechnical Director Commander

NO"TICES

DISTRIBUTION: Fromrrey distribution v! this report has bwNh maleo by ARI|. Plem edrs oreseSnldencJonCIRVin distribution of reports to: U. , At"rmy e&M" Institute #Of theIiellthsioad w~do"ATTN: PERI.'rP, 5001 Eiverho.mf Avenue, Aklexadris, Viro~nle 22333.

: This rep~ort! ry be destroyed whion Is Is re woomood. Pwem d3 not rmrn it tothe U. 8J Art" flieporch Inotitutip for' the ShipMelrof on SojW~e kN~

STheindirip in this report we not to be ownodtm~ aI on oflfoW Deo of the Amw pftmi,unlessl so desipwm by other' authorized doculments.


3/74

UncIlasaif LedSECURITY CLASSIFICATION Of THIS PAGE CMwn Date A;"..*REPORT DOCUME14TATION PAGE BEFOE RCMPLTINFORK1. RIEPORT NUMBER 7 jVACslow No 3. FRECIPIENT'S CATALOG UMummaResaearrch Reot_1237 f&~l/e _______TiZ11.E f-'- 2-3 S XEOF REPORT Is PIEROD COVEREtD(~ MN O9SSS IN INELGNE#ALYSIS: /1~ia ,i

RAE IM RVSISPW S. PERFORMING oRe. ReponT MUMMER

R~brtV.--ttor) &.rirtifle A- 0rntqOW* MI 3-478-C-2 44/as John R. 2=QV. P~RI110IP"MM=hTYATbON MAM'NOADDRIKIS POGA SLEMEN9T. PROJECT. TASKOperating Systems, Inc.!21031 Ven~tura Boulevard"N 3801Woodland Hills, CA 9136411. COHTRtOLL.INGOFPFICEMAI4E AI~IADDRIESS ILI S Army Intelligence %Security ComumandIASYS-R, Arlington Hall Station MS"1AEAxlington, VA 22312 6ol~~~~rtING(0Eef 14ANSI & ADORES~rt Afle*d Now Cm~oerlind 0ff..e) is. SECURITY CLASS. (of Oft tiPe)US Army Research institute for the Behavioral &UnclassifiedSocial Sciences I& D~ IS 1PR DNERI-OS SCMSSLICATI614/00VRAING5001 Eisenhower Avenuea, Alexandria, VA 22333 _____________

Is6. INSlRIUTioN STA TEMENT (of Of* RAPMf)Approved for public release; distribution ijnlimlted

*?. DIST IP4UTION ST AT 914 NT (of A.o"street a~t~rd n EI.E 4",f Offm sit 600 R*pmt

IS.SUPLEENARYNOESThis effort was Jointly sponsored by the US Army Intelli-gence & Security Coumand (USAINSCcH) and the Army Research Institute (ARI).Dr. Edgar M. Johnson (ARI) and Mr. Philip Lavigne (INS COM) provided programdirection; Dr. Stanley M. Halpin (ARI), MAJ William Franklin (INSCOM) andk ~ ~~Dr-.uth H_ . Phgt a (A T mntnx. I-achn 4in19. Key WORDS (Canthwe !" reV 61* If flaewy aw l fd.mtI 5. S' biAnalysisIntelligenceImageryAnalytical Procedures

USQAfrMACr(bW? t mm wW ,.. - 4P6U"0 DW b' This report provides anoverview of the results of %. study entitled h'Investigation of Method~ologiesand Techniques for Intelligence Analysis.4 The goal of this study was todevelop a framework for understanding human processes in intelligence analysis.An understanding of thaese processes will be useful during the development orevaluation of training procedures, doctrine, an d system requiremants forautomated support to analyets._

DO S Uuro o NEOEECLASSIFIEDSKTLU AMPIPCATION Olt THIS PASE (10 Dael 1iii.weE)


4/74

UNCUV4TY CLAWsIPCA'riovs OF THIS PAGR(Whm Date m0*

(cont). D-A@ '95iThe approach token in this project was a parallel development of intel-

ligence production models and a cognitive model of intelligence analysis. TheImagery Intelligence (IMINT) Production Model (Au iutwear.Ji Repodeveloped under this project and an existiqg Signals Intelligence (SIGINT) pro-duction model were used as the basis for developing a generic model of intel-ligence production, described in this Overview. The generic production modelincludes a description of the hierarchical f'1Vow of intelligence data, tasking,etc., as well as a listhig of critical variables influenicing analysis. Adetailed review of the psychological literature described in u*mI Trutzuvu )4tpwrt 445,)combined with the production models, led to development of a de-scriptive model of cognitive processes described in this 4,erview. This over-view also explores some of the implications of the cognitive model for trainingdevelopments, system development, and organizational cc trol of intelligenceproduction. The cognitive model will be extended and these implicationa willbe examined in greater detail during Phase 11 of this project.

JUnclassifiedon ~ CASFC1SNO HI A~W~ aaEtad

B


5/74

Research Report 1237

HUMAN PROCESSES IN INTELLIGENCE ANALYSIS:PHASE I OVERVIEW

Robert V. ItatterChristine A. MontgomeryJohn R. Thompson

SibmIttWd by:Edger M.Johnson, Chief

HUMAN FACTORS TECHNICAL AREA

Appovad By:Jaime D. Baker, Acting Directo"ORGANIZATIONS AND SYSTEMSRESEARCH LABORATORY

U.S. ARMY RESEARCH ,NSTITUTE FO R TH E BEHAVIORAL AND SOCIAL SCIENCES5001 Eisenhower Avenue. Alixandria, Virginia 22333,Office, Deputy Chief of Staff for PersonnelDepartnent of the Army

Decemb:er 1979

vfproIct Number IntIllignce System20163743A774

-OF orel for publi 'ssm~g dismlastni sdm*h.us11

A% SII


6/74

AR I Research Reports and Technioral Reports are intended for sponsors ofR& D tasks and for other research and military agencies. Any findings readyfor implementation at the tire of publication are presented in the lost partof the Brief. Upon completion of a major phase of the task, formal recorn-rnendatlons for official acttor, normally are conveyed to appropriate militaryagencies kv briefing or Disposition Form.

:iv


7/74

FOREWORD

Intelligence collection systems have proliferated over the past severalyears, increasing in complexity and In volume of output. However, there hasbeun no corresponding Improvement in the ability of intelligence personnel toanalyze this flood of data. US Army Intelligence and Security Command (INSCOM)studies and Army Research Institute (ARI) research indicate that Improved supportto and training of analysts are necessary to effectively utilize the increasedcollection capability and satisfy increasing demands for intelligence withincurrent personnel constraints. INSCOM and MI therefore initiated a joint researchprogram to provide improved support to the intelligence analyst. During earlydiscussions of the issues, it became clear that any procedural, training, organi-zational, or system changes to support analysis will be effective only if basedupon a detailed understanding of the analysts' role, methods, and thought processesin intelligence production. The first need was to evaluate and describe the humananalytic processes underlying intelligence analysis, synthesis, and production.This report provides an overview of the results of this research.

The approach taken in this project was to examine the role and activitietiof various types of intelligence anplysts, and to develop a descriptive modelof the cognitive processes involved in analysis. This model derives in largepart from current psychological literature and does not provide a new statementof psychological principles. However, these principles are put in a new context,intelligence analysis. This report describes the generalized intelligence produc-tion model and the cognitive model. Separate reports describe in detail a modelof the production of imagery intelligence (ARI Research Report 1210)-and-O A . Ipsychological literature underlying the cognitive model (ARI Technical Report 445)_/

The research was accomplished by a government-contractor team under contractMDA 903-78-C-2044 and was monitored jointly by INSCOM and AI. Continuous inter-action and collaboration of personnel from Operating Systems, Inc., INSCOM, andARI insured a multidisciplinary approach to this research.

This report and the others from this project provide a framework for detailedexamination of training support and system support requirements in intelligenceanalysis. These reports should be very useful during the development or evalua-tion of training procedures or materials, aralytic procedures, doctrine, andsystem requirements for automated support to analysts.

Di. VOSEV DNEZ MG WILLIAM I. ROLYArkc ical D rector for AuI :nd Commander,'US ArmyThf Psychologist, US Army Intelligence and Security".ommand


8/74

ACKNOWLEDGEMENTS

This study was conducted with the assistance and advice of a Study Advi-sory Group (SAG). The contribution of the individual SAG participants andthe cooperation and support of the agencies represented by these individualsgreatly benefited the accomplishment of study objectives. In addition tothose named below, several INSCOM personnel participanted in the four SAGmeetings, several other intelligence activities provided access to analystpersonnel, and 117 individuals contributed as much as 6 to 8 hours of theirtime during interviews and discussions. This high level of participation andsupport across the intelligence community is gratefully acknowledged.

SAG PrrticipantsG. A. Harvey, Jr . INSCOM SAG ChairE. M. Johnson AR I SAG Co-ChairH. Woodall ODUSA-ORH. W. Taylor OACSI DAC. L. J. Legere USAISDJ. E. VanSant TRADOCCOL S. T. Weinstein XVIII CorpsJ. Ribholz NSAA. Heidig DIALTC C. H. Keller CACDAP. E. Lavigne INSCOMMAJ W. A. Franklin INSCOMCPT M. Hainline USAICSS. M. Halpin AR IR. H. Phelps ARI

Vi i


9/74

Human Processea in Intelligence Analysis: Phase I Overview

BRIEFRequirement:

To develop a framework fo r describing and underatanding the process ofintelligence analysis as performed by analysts workitig with either single-source or multi-cource data.Approach:

A series of structured interviews was conducted with both national andstrategic level intelligence personnel experienced in signal, imagery, andall-source int1lligence processin& and production. Interviews with devel-opers oi training materials and observation of classroom instruction andfield exercises provided additional information on individual and groupresponsibilities. Data obtained from interviews and observations were3upplemented with documentation and reports based on INSCON and AR I research.At the same time the research literature on cngnitive information processingwas reviewed. A descriptive model of the cognitive processes underlyingintelligence analysis was developed baced on the general principles derivedfrom the literature review and the intarviews with intelligence personnel.Product:

The examination of intelligence analymis identified enviromental and-ndividual variables as well as underlying cognitive processes which con-tribute to the quality of intelligence. The model describing the cognitiveprocesses of intelligence analysis is expected to serve as a descriptiveframework for understanding all levels and disciplines of intelligenceanalysis. One of the main findings is that intelligence analysis is aninternal, concept-driven activity rather than an external, data-drivenactivity. The present report is an overview of the project, including thecognitive model. A process model of imagery intelligence and a detailedliterature review have been published as separate reports.

The findings and descriptive cognitive model could potentially contrib-ute to the improvement of the quality of intelligence products by providinga better understanding of the thinking, analyzing, and memory processes ofintelligence personnel. The model can be used to study management andsystem functions as well as to guide future training and system develop-mi Its. The results will be incorporated in the development of functionalsi cifications for the All Source Analysis System.

tx


10/74

CONTENTSi1. INTRODUCTION AND OVERVIEW ........ 0........... ... ....... 1-1

1.1 Basic Objectives ..................................................................................... 1-1S1.2 Overview of Topics ..... 0.0,...000**..I...0*.. ..... 0..*. ..0.... 1-2S1.8 What is Intelligence Analysis? ............................................................... 1-2S1.4 The Core Processes cf Intelligence Analysis ......................................... 1-3S1.5 Study Approach .. ................................................................................... 1-3S1,6 Conclusions and Implcations for the Cognitive Model ............................ 1-10

i2. FRAMEWORK FOR INTELLIGENCE ANALYSIS COMP:ARISONS ............................. 2-1!2.1 Organizational Structure of Intelligence Analysis Tasks ......................... 2-2S2.2 Communication Factors ........................................................................... 2-2

3 . GENERIC INTELLIGENCE PRODUCT.ION MODEL ............... ........................... 3-1! 8~~~~~.1esource Managoment ........ ....... .................................... 3-23 .2 Adiaptetlon ...................................................................... 0...................... 3-23.3 Intorpretatlon ......... 6.... 0........ 0........0...... 40-1.... 0.......*6.... 0........................... 3-44, ENVIRONMENTAL VARIABLES .................................................... 4-1

4.1 Assigned Domain ...................................................... ................... 4-14.2 Management and Tasking ...................................................................... 4-14.3 Client Orientations ................................................................ ................ 4-14.4 System SUplpits ................................................................ *.0*.......... 4-14.6 Knowledge I t .................................................................................... 4-24.6 Judgmental Criteria ...................................... ........................... 4-84.7 Information Acquisition.............................. ............................................ 4.3S4.8 Data Types Familllsdty/sk~lls ................................................... ............. 4-3S6~~~~~~~. ANALYST VARIABLES ............................... .. ......... ........................ 1S6. 1 Knowledge .......................................... ................-......... .......... 45-15.2 Skills ....................... *..e... 0..............0........... ......................................... .. 5-28.3 Memory Management Orientation ............ .0*... ............ ..... 0................... 5-38.4 Peroonallty Attributes of the Ideal Analyst ....... .................................... 5-6

8. TH E COGNITIVE MODEL ............................................ ..... ............ 0.....0........... 8-18.1 Overview .................... *....*................. ......... @.......... .. ..*.................... 6-16.2 Information Contents Modification Cycle ........................... . .*.........*....... 6-38.3 Structure of Cognitive Capacil es ......................................................... 6-86.4 Elemental Dynamic Proceesses ................................................. 6-14

7. THE COGNITIVE MODEL IN THE ANALYTIC WORK SETTING ............................ 7-17.1 Special Importr~nce of Memory ................................. 0........0..................... 7-17.2 Awareness of Memory Function ............................................... . .......... 7-17 3 Me mory A ide a n d Me mo ry Loa d 0............ 06.... 0 000....... . . .. . . .. . .. . . .. . .... 7 - 17.4 Decision Pressure ...................... 0........................................ 7-27.6 Memory Resource Management ............................................................. 7-47.8 Team Memory ......................... ,0 ......... f*.................................-47.7 System Aggregation Models for Datea nd Concepts .............................. 7-6

xi


11/74

8. IMPLICATIO!"S OF FINDINGS ................................... ................... S-18.1 Analytic Accuracy ..................... .......... ......... .................... .... .8.2 Management of Analytic Work ............................................................. 8-18.3 System Supports for Analytic Functions ......................... ............ 8-28.4 Performance Feedback Arrangements .... 8-28.5 Training for Analytic Work ..... ....................................................... 8-38.8 Selection and career management ................................... ....... s........ 8-3

9. DISTRIBUTION ............................................................ 9.19.

xti

2:A


12/74

HUMAN PROCEASSES IN INTELULGECE ANALYSISs PHASE I OVERVIEWRobbwt V. KatterChristineA. MontgomeryJohn A. Thompson

Operating Systems, Inc.

ABSTRACTA project was undertaken to construct a general descriptive model of theoognitive avities (mental processes) underlying the activities of Intell-goene analysis. The apprmach used to develop the cognitive model comn-blned avalaeble Information on the way In which intelligence analysis Isperformed In actual work settings with avallahle research findings In cog-nitive psychology. The first step was to Investigate analytical processings currently practiced in two types of single-source analysis, and subse-quentty to generalize to multi-aource analysis. The initial Interview andobservation field study Investigated signal Intelligence (SIGINT). A secondfield study investigated imagery Intelligence (IMINT) and resulted in amodel of the diresUy observable activities of single-source IMINT produc-tion. A questionnaire interview guide was developed and used In deeperstudy of the two single-source disciplines and multi-source productionactivities. A concomitant task reviewed literature from cognitive psychol-ogy for applicability to the study. A cognitive model was developed, andIntelligence analysis activities were studied from the point of view of themodel. A main finding is that effective Intelligence analysis Is a concept-driven activity rather than a data-driven one. Project results have Impli-cations for improvements In a number of areas of Intelligence work.

1. INTRODUCTION AND OVERVIEW report [ Imagery Intelligence (IMINT)1.1 Bernie Objectives Production Model ] was produced bythe IMTIA project. PubllcAtions used inThe study entitled lnvestigatPmn of the project are Included In the refer-Methodologies and Techniques for Intel- ences.Igence Analysis (IMTIA) has been aimed T1e IMTIA project alms to Improveat constructing a model of the cognitive understanding of the cognitive activi-processes underlying Intelligence ties underlying Intelligence processinganalysis activities. The present report and production. While the IntelligenceprovIdes a management overview of theproject and Its results. Another report rea soniifed andi under-seCgiiePocse nItl/ reasonably well-defined and under-geoee Anieyls oDescriptive Model stood, the activities that go on In theanid Review of the Literature I provides head of an Intelligence analyst are not.an expanded version of Section 8 'of These Internal activities have beenah prexpnd veprsdio sctssiong te mainly a 'black box', where only Inputsthe present report, discussing the and outputs can be observed andtechnical literature basis for the cogni-tive model In terms of Interest to social described, and the unobservablesclc- triots and technicians. A third processes Internal to the black box are- tta tcia.unknown. The IMTIA project is directed

- ~-1


13/74

at filling this void in tho overall under- Intelligence analysis and the formulationstandlnj of intelligence processing and of operating policies and procedures;production. The basic objective of the Providing criteria for the developmentstody was to develop a oognitive model and evaluation of advanzed systemthat would provide a framework for the designs In support of Intelligencedescription of the mental processes analysts; Providing criteria for improvingused In intelligence ana,/sls. system arrangements for performance1.2 Overview of Topics lfoeeback; Facilitating efforts to

develop and evaluate training programsSection 1 summarizes the development and training doctrine; and, Providingapproach for the model and outlines the assistance In personnel selection,main results and conclusions. Section 2 motivation, and career planning.through 6 are based on Interviews andobservations in the field. Section 2describes the organizational structure The definition of Intelligence ana!ysisof tho intelligence cycle and treats presented in the Glossary of *ntelll-Information flow within that framework. gaene Terms and Definitions [seeSe'otion 3 presents a generic model for references] reads as follows: "A pro-single-source and multi-source InteAli- cess In the production step of thegence production, Jescribes the core Inteligence cycle In which Intelli-InforVton processing functions of gence Information Is subjncted toresource management, adaptation, and systematic examination In order toInterpretation, and outlines task Identify significant facts and deriveprocesses common to Intelligence conclusions therefrom."analysis. Sections 4 and 5 describe The problem with this definition is thatenvironmental and individual variables The terbs, when used to describe--that affect Intelligence analysis activ- tetrs hnue odsrb"tat. processes inside the Oblack box", have

no frame )f reference: What doesSection 6 presents the cognitive model "systematic' mean? What are the sub-developed In the course of the study. processes In "examination"? WhatThe model Is synthesized from a review about "Identification'? Osignificant"?of recent cognitive psychology litera- 'facts'? Oderive'? "conclusions"?ture. This major section describes the Althouqh the cognitive model now pro-structure of cognitive capacities, the vides a framework for answering thesebasic dynamic cognitive processes, and questions, no such model existed at thethe cyclic modification of memory con- outset of this study.tents. Section 7 describes aspects of/ ~analytic work In the light of the cognl- Inodrtpusehsreac,several operating definitions werethv model, and is based on field obser- adopted. A basic defnition was thatvations made from tha point of view of Inted.igence analysis is what Intelli-the cognitive model as It wasdeveloped. The final section of the gence analysts do -- an understandingreport, Section 8, presents implications of the processes of Intellgenceof the cognitive model for a number of analysis an d the cognltve processesareas. Those include: Improving the underlying them must cm e from theaccuracy and quality of analytic pro- analysts themselves, faon observingducts; Improving the management of what they do and asking them about

- 1-2 -

-~ .~. .-. ~-.-.--.---


14/74

A subsidiary definition was that the support core Information process-performance of analytical and judgmen- Ing functions fIr intelligencetal activities Involving complex cogni- analysis.tive processes is not limited to person- 2. Second, t was assumed that thenel traditionally called analysts. identification of these commonRather -- for the purposes of this study analytical task processes was a-- Intelligence analysis was defined to keytidty t isk ore was okey to identtfyng the core Infer-Include a spectrum of analytical and mron processng functions ofJudgmental activities Involved in the m Impoce ointionce-nost Importance to intelligenceprocessing and production of intelli- analysis. This was viewed as angence, where Individuals assigned topartculr Wenthi proessmayImportant stop In developing a cog-particular roles in this process may nitive model that could provide anspend more or less time In analyticalactivity depending on their assigned especially productive framework forroles. describing Intelligence analysis.1.5 Btudy ApproachFor example, signal Intelligence (SIGINT)voice intercept operators clearly per- The approach to identifying the set ofform some analytical activity in search- common analytical task processes andbIg the signal environment; however, developing a generic cognitive modelmuch of their duty time Is devoted to based on these was first to Investigaterather rigidly specified collection activi- analytical processing as currently prac-ties, Involving a minimum of analytical tr ticed In two types of single-sourcejudgmental processing. On the ethes analysis, and then to generalize thehand, Image Interpreters (IMINT findings to the multi-source analysisanalysts) In a national strategic facility environment where possible. As shownare more continuously Involved in In Figure 1-1, the assumptions dis-analytical activity In searching imagery cussed above and the Initial conceptsand must deal with more complex data regarding the nature of the analyticpatterns. In contrast to SIGINT and task converged to produce the cogni-IMINT specialists, multi-source analysts tive model described In Section 6. Thedevote the largest proportion of their work represented in the top half of Fig-time to analytical activity and deal with ure 1-1 was a data-gathering, analysis,cxmplex data patterns representing and synthesis effort directed at defin-aggregations of single-source data. ing the process of Intelligence analysis

1,4 The Core Processes of ,,tU- as performed by its practitioners. Thework represented In the bottom half of*efl~e AnaIysis the figure was directed at defining theTwo assumptions about Intelligence underlying processes of intelligenceanalysis underlie this researcht analysis in terms of cognitive functions.The cognitive model thus Is a blending1. First, it was assumed that a se t of of the Information gathered and gen-common analytical task pocesses erated In answering these two ques-exists that crosscuts the various tigrs.

Intelligence disciplines such asSIGINT, IMINT, and HUMINT (intelli-gence from human sources). Thesecommon analytical task processes

- 1-3 -


15/74

WHAT 33

WHAT, AR t1

Mode)A duct JoodelU-OIdenifythoeGIN tas prceses ainc nrodtefomprbepouto"oewhih ppar ov ahig aaltial xite fr IIN, hemodldsrieanddgmetal content. _hese were in [ maryInTeIgenc (IMIN)Po-

model~~~~PeeserepareofISOM cnen. h sletd roese wr

[. .I1INtudProduction MPodeluTctinc& comparveopednwthoihla SIGNT taskIdentify proIN akpocse icese inoomprdbert provideian inidil

Invetigaions te SIINT rodudefnIntliong ofgcomnayil analyticalntasknmdel~~~~~~~~ preareo-NC otn.Te eetdpoesswr

SIIN _rdcto iomae ihsmia ITts


16/74

processes that crosscut the Intelll- e Motivatlonal Aspects -gence disciplines, respondent's views of career path1.6.3 Preparaeionof Quetio~rneire or models, personal goals and expec-l or Coletion tations, goal attainment criteria,relevant skills and knowledgeThe bes:s for development of the IMINT base development.production model was the Intelligence 1.A t V anr A Inter-processing behavior of IMINT analysts. viewSIn order to systematically cc,'sct infor-mation on IMINT anilyass and tQ develop The questionnaire was used as a basismore extensive Intormatlor, c SIGINT for cenductiog the Interviews listed Inanalyals as well, a detailed question- Figure 1-2. Although a number ofnasre was prepared for Intealewing relevant Army and DoD manuals andIntelligence personnel In both single- reports also proviaed Important basicsource and multi-source Intelligence dita for this study (see references),production activities, the richest source of Information wasThe questionnaire covered the following the oommunit, of analysts interviewed&rags of questions for each interview and observed as they carried out ths!rrespondent:e Dormain Definition- respondent's Of pIarticilar Interest are the termsDomnt orientition,m analysts use to refer to themselves andclienttheir work. They often se e themselvesreeponsibllity, Input types, Infor- te "deTe es hfte rsee the rmation sources, knowledige base, &a "detectives," "historians;" theirpatdontsoures. k, work Is Ilke "solving a mystery," a "puz-product types. zle," requiring "deduction," "imagina-

s Percepts and Concepts - tion," and "complete concentration."respondent's perceptual modes The guiding principle is not only to lookand skills, mental Imagery types, and to listen but, -more Importantly, todifficult/easy tasks. think.* System/Procedural Contexts - These concepts provided insights into

patterns of communication activi- the Identification of the core analyticties, management and quality con- functions and the cognitive mechanismstrol arrangements, In rpretive underlying them. The evolving cognitiveaids and references, procedural model was updated and refined In termsflexlbilltles experienced by of data that Indicated significant vari-respondent In work setting. ables in the analytical environment-respondent's (described In detail In Section 4.1), Ino Decision Procsses - res individual training, experiencq, andview of work setting situation with knowledge base deveiopment (Sectionregard to conditions for making 4.2), and In key processes In theInterpretation decisions, Interpre- anl on (ectpon6).tation decision controls, adapta- analyst's cognition (Section 6)."tion decision mechanisms used. 1.6.6 Selectio'n of Aneiytijal Process-and prevailing adaptation arrange- Ing Functions for In-opth Studyments, In both the SIGINT and IMINT studies, aset of analytic processes--Identified as

- 1-5-


17/74

MONTH I OfteANZAT1101 N0. INTERVIEWS/SPECIALTIESARMY INTEWQEiICE AN D SECURITY COMMAND (INSCOM) 2/S10INTNOVM~N NATIONAL SECURITY AGENCEY (NSA) 6/SIGINT, 2/IMINT

DECEMBNER ca /MTNATIONAL PHOTOGRAPHIC INTE~RPRETATION CENTER (NPIC) 6/IMINT

626 Ml GP, MIBARS, FT. BRAGG 0/IMINTARMY/A? EXERCISE, SHAW AFB 12/IMINT3426 TECH TRAINIWG, OFFUTT AFS G/IMINTMARCH USAICS, FT . HUACIIUCA 13/IMINT, 2/SIGINTINSCOM 3/IMINTNPIC I /IMINT

APRIL FT. DEVENS 1CjSIGMtT

JUNE BLUE A-AG. F-QLIN AFS S/MULTI-SOURCE

FORSCOM EXERCICE, FT. A.P. HILL I G/bULTI-SOUACECPAR, FT. MEADE Q/SIGINT

AUGUST DETERMINED EAGLE FTX, FT. BUISS 12/MUL rI-SOURCE

SIGINT a 20 INTERVIEWSTOTALSI IMINT o 5a INTERVIEWS 1 17 INTERVIEWS TO DATEMULTI-SOUJRCE 338 INTERVIEWS

Figure 1-2. Summery of Sites V~i'uted for Data Gathering

-1-8-


18/74

rserving functions common across the For purposes of comparing SIGINT anddisciplines of SIGINT, IMINT, HUMINT, IMINT, the following analytic processesand multi-source Intelligence-- was commoa to both were selected:selected for further analysis andcross-discipline comparison. * Search0 Target DetectionThe following criteria were used toselect analytic processes Incorporating 9 Target Identificationcommon functions: e Target Devolopment* Analytical orientation - whether

the process Is mainly analytical orprimarily control an d management a Functiorial Analysisoriented. * Complex Studies (e.g., OB

e Generality - whether the process Analysis)appears general o Intelligenceanalysis or specific to the particu- * Asociated Reporting Processeslar discipline of SIGINT or IMINT. 1.5.6 Comparisonof SIGINT and IMINT

e Interpretive Complexity - Analysiswhether the process Involves Using the common task processessparse, possibly conflicting, data selected above as a framework, com-that Is subject to many potential parisons were made between SIGINTinterpretations, or detailed and and IMINT along the following dimen-unambiguous data whose interpre- slons, which are described in detail intation Is relatively straightforward. Sections 2 through 6:

e Level of Aggregation - whether e Strategic versus tactical intelli-the r'ocass Involves discrete, gence production.Iow..level data patterns or a more * Environmental and individual varn-abstract level of Information that e sdhas been aggregated one or moretimes (for example, SIGINT target - Type of data used.detection veratis multi-sourcefusion). - Work setting.

Figure 1-3 shows an Initial selection - Client mission.chart for IMINT processes in terms of - Training, experience, andthese criteria (abbreviated to ANAL, knowledge base developmentGEN, IC , and LOA, respectively). The L, and maintenance.M, and H labels in the figure Indicatethe evaluation of each process as rela-tively low, medium, or high with respect e Use of complex conceptual modelsto the given criterion. Processes that in analysis.were evaluated as low In analyticalorientation or generality across all thrae SIGINT and IMINT are generally per-ceived as entirely different because ofrocessing phases were not selected.The processes selected are boxed Inthe figure.

- 1-7 -


19/74

inMr PHASE UrC0 PHASIE ThIRD PHASEANALYTIC ACTIVITY PiOCMBSM PROCESSING PROCESGINGKiRcTMEN ANAL LOAJ ICNE N WAL LOAJ ICMENAA.

ITERPRETATION

Search L HL L MH"M N1 H HHHargetDetectlon L II H H M'M H H H H " HTarget Identification L M H H M M H H H H H HTarget Quantification L M H H M M H H H H H HUnusualnessAnalysls L M H H M M H H H H H HFunction Analysis H H H HComplex Studles I HH HPhotogrammotric Sciences L L L HREPORTING

FIrst Phase Reporting L M H HSecond Phase Reporting M U H HThird Phase Reporting H H H HCUPPORT ACTIVITIESCollection Coordination H H H L H tL _UL _ _H_ . . LAll Source Data Review & Reduction_ H H H I HH H H H H H .L H

Hlstory of Coverage Maintenance L L L H I LL L L L L HTargetDevelopment [I H H H H H H H H Hi H HInterpretation Key Develt anent H H L H H H L H H H, L HKimwledeBase&IIIlUDevopme H H L N H H L H H H L HLA-Level ofAggegation L o1 fic - iterpetly Complexiy MAmou A t Of ftovtWyGenerality HHigh

ANAL - Analytic Orientation

FIgure 1-3. Analytic task processes selected for additionalstudy are enclosed. Any activity scoring low in GEN or ANAL forthe three processing phases was eliminated for further study.the dissimilar rmodes of data capture depicts a relation or relations betweenand resulting perceptual differences two or more entities, where an entity(hearing signals and viewing signal- may be an object, an individual, or ananalysis displays versus viewing eve it.terrain-based Imagery). In fact, some Fo r example, MINT analysts may haverather striking parallels emerged from conceptual model of a surface to airthe comparison of SIGINT and IMINT missile, In which the model depicts aanalysis. The most significant of functional Interdependency between athese was the use of complex con- missile and Its transporter. If they thenceptual models in analytical work. see a missile transporter with no missileA conceptual model consists of a pat- on it, they may search the surroundingtern of generalizations about a given area for places of storage or conceal-category or range of experience, which ment for the missile. Examples of more

- 1-8-Ii-,- . -:.-" -:'?l '-"': : .'r:- - - ,'-T.--" '--,- '- . .. -.--' . . . . .-


20/74

complex conceptual models are the While IMINT analysts are able to spendideas of a communications net, an Order time in building and maintaining theof Battle, a Table of Organization and store of conceptual models that formsEquipment, the general relationships the Internal knowledge base, SIGINTbetween terrain physical characterls- analysts on an Inactive mission aretics and mob':Ity characteristics, a col- more concerned with not losinglection system, etc. language or morse proficiency, ratherthan with Increasing or maintaining gen-Another important parallel between SIG- oral or military knowledge.INT and IMINT analysis emerged fromwhat analysts saw as the key attri- Both SIGINT and IMINT analysts findbutes of at good analyst. Both SIGINT assignments to Inactive missions in theand IMINT analysts stated that a good tactical environment frustratinganalyst Is a "detective", that Is, one because they like their analytical workwho discovers what Is concealed or and know that lack of practice leads totends to elude observation, one who is loss of proficiency. The SIGINT analystssolving a mystery or a puzzle. The have more to lose In such a situation,effective analyst must have 0imaglna- since data-captUre mode knowledge"tkN In order to generate good (language, morse), as well as generalhypothes, s about what Is happening and military knowledge, will Inevitablyand what beight be concealed, and to do degrade.some "detective woric" to collect evi- Similarly, although both SIGINT anddance that confirms or refutes these IMIhT analysts prefer analysis tohypotheses. It Is Important to "be a management, SIGINT analysts whohistorian" because one must know what moved to management positions wherehas happened In the past In order to their voice or morse proficiency was notunderstand what Is happening In the used expressed the most frustrationpresent and what Is likely to happun In because of the sub'tantl amount oftise future. time they haa devoted to their early,The major differences between SIGINT extensive, career training.and IMINT analysis are attributable to In eneral, the similarities between SIG-differences In modes of data capture INT and IMINT analysis far outweigh theand resulting different perceptual iNanIMNaalssfrotegthrequirements. Several of the SIGINT specific differences attributable to dif-spe requirements. 8 greaal ofheSINT ferent data-capture modes and, conse-speciaftce drequireaatgreatrdealn quently, to different modes of percep-specific data-capture training tion.(language, morse) In addition to generaltraining In military science (and other 1.5.7 Extension of the Comparisontoareas discussed In Section 5). IMINT HUMINT and Multi-Source Analysisanalysts on an inactive mission may Alhough a detailed, intensive Investlga-devote off-duty time and Idle time on thou HMtas ntewsthintescopedutytocreaingthei genral tlon of HUMINT was not within the scopeduty to increasing their general of the study, some observations ofknowledge about the geography and HUMINT analysts at work were madecultural background of theoret tp %os- and several discussions were held with

IPW (Interrogation of Prisoners of War)military equipment likely to appear lr analysts. For the most part, It appearsImagery generated during exercises. that the generalizations about SIGINT

- 1-9-4

1i-~ii


21/74

language specialists made In the o Ability to acquire anti comprehendpreceding section also apply to HUMINT complex conceptual models.analysts. A major difference Is thegreater importance of special verbal now tonertean eluskills for HUMINT analysts, who spend (hypotheses).many hours interviewing in a foreignlanguage. Anoier difference is the o Knowledge, consisting of a generalrelatively greater Importance, for mental Inventory of available con-HUMINT analysts, of access to multi- ceptual models, Including mentalsource Information, which provides them "indexg to locations of models inwith the neoessary knowledge base for external memory (in the form ofhelping to elicit Information from per- colleagues, files, references, aids,sons being interviewed. etc.).Extending the comparison to multi- o Accurate, reliable memory func-source analysis, the major differences tioning.between single-source and multi-sourcea i Special Interests and orientationsSanalysisnvolve: toward conceptual modeling as an

o higher levels of aggregationof the activity in its own right, Including* Information to be analyzed in explicit recognition of models asmulti-source, and to some extent, mental phenomenon, and selectiveattention and selective memory foro greater Interpretive complexity forceti knd of ocpuamulti-source because of the rela- models.tively greater number of concep-

tual models to be considered. e Mental flexibility and ability to1.6 Concluskms an d Implications fo r lear from past experience.

the Cognitive Model The use of such models is also depen-i ~~The clear implicarsution th~ study le vaibe imlpIcting~ th analytic pro-Th e most telling result of thle study is dent upon environmental or work settleaSthe clear Implication that Intelligence vaibeImctn h nlycpr'analysis Is conceptually driven as ceas, sdch as:opposed to data driven. What is critical 9 Client orientations (support of tac-is not Just the data collected, but also tical versus strategic command,what is added to those data In Inter- Indications and Warning, missionpretIno them via conceptual models in opa, plans, logistics).the analyst's store of knowledge. Prd cin m ag et a d t sk Production management end task-The most Important parallel between Ing (mission, area of interest, pro-SIGINT, IMINT, HUMINT, and multi-source duct accounting, tasking agencies,analysis Involves the use of a large work breakdcwn structure).variety of complex conceptual modelsIn analytic activities. a Information acquisition (uoliectionplan, collection gaps, predictabil-The core functions of Intelligence ity).analysis Involve the use of complex a Domain (geographic areas,conceptual models. In turn, the use of sociocultural characteristics, ter-such models is dependent upon Indivi- rin features, weather,dual variables such as :

- 1-101


22/74

economics).* Date types (sensor types,observers, filtered data, analytical

results, expectations).e Judgmental factors (priorities, timoavailable, communicationbandwidth, significance).e System supports (displays, databases, colleagues, references,computational aids).

Because both Individual an d environ-mental variables affect the analyst'scognitive performance, the model ofcognitive proUesses underlying intelli-gence analysis must show sensitivity tothese v.riables.

S~-1-11 -

" [_ _'_ 'S''--" N.... . . ... .... ... .. . ..


23/74

TyPse OF 1FOAr UrTTINOSCOLLECTION/SINGLE MULTI- CL/RN?SOURCE SOURCe PRODUCT

PRODUCTION FUSION PACKA GINGajb VOICK IAIG

INTELLIGENCE rPSC nttINFORMATION IR H UM I IIT or.71AOTOT - IP/ CDR.I.COMMUNICATION.COMMUNICA TIVErUODSLSI/ I ,h'.st I .,

CONCEPTUAL /Contingency Plan.s 1K .. 1 LMODzELING / Intellig.ec. ..dut.on/ CapIaiih !.J/ SiuationaL Model. I IMdel

FORMAL INFORMATION LOW I ICOLLECTION Z ' ,Req.,remon.tIj/ I ' 1I I. Tmasking _ _FIRST PHASE _________,___,___I_!___________:SECOND PHASE FRaw Dat"THIRD PHASE ,

ANALYST J05 ROTrATIONFigure 2-1." Structure of Relations Between Analytic SpecialWe.

2. FRAMEWORK FOR INTELLIGENCE particular context within which the workANALYSIS COMPARISONS setting resides. Analysis of the overallThe main dimensions for comparing Intel- organization of Intelligence productionactivities In such terms led to thegence analysis specialities arethe multi-dimensionalreflected In the work settings In which structure shown In Figure 2-1, whichthe ipecialitles are performed. These depicts work settings for various typesimensions Include differences in deitwokstngfrvrostysaimeiounts andcmdeso dataer so b of SIGINT, IMINT, and HUMINT process-mounts and modes of data to be I g n h i e ai n t o k s ti ganalyzed, assigned domain from which ing, and their relation to work settingsdata are generated, kinds of analytic In which higher levels of Intelligencetasks to be performed on the data, the data processlng are accomplished.general range and conceptual level of The bulk of Figure 2-1 consists of fourconcerns that are to be brought to bear Interconnected sets of columns depict-In considering the data, and the Ing the overall Intelligence system. At

- 2-1 -


24/74

ILI

the left of the figure Is shwn the Intel- settings are also linked by four kinds ofligence Information cycle, which con- communication factorsis sto of a generalized sequence of fourInformation processing phases that * Communication models.occur throughout the syste" at most a Formal Information flow arrange-levlls and for meot modes of data: ments.

9 Collsotion of data that is Indica- a Informal communications.thiv of capabilities or Intentions. a Job rotation.e 11irst phase exploitation at data Each of these are considered below.(high threat and limited responsetime window). 2.2.1 Communication Models

e Second phase exol cation of A communication model Is a commondata (for current 3rations and conceptual model shared by parties whosituation assessment), use It as a means of communicatiom,.SThird phase exploitation of data Such shared conceptual models linkSsupport basic lntelligencs intelligence work settings. As definedearlier, a conceptual model consists ofneeds and special requirements, a pattern of generalizations about aand review of conceptual modelsused to Interpret data. given category or range of experience,which depicts a relation or relations

2.1 Organizational Structure of Intel- between two or more entities, where anIlgence Analysis Tasks entity may be an object, an Individual,The main types of settings for Intelll- or an event. Shared onceptual modeisare the basis for common understand-gence analysis are portrayed across insfmayutlcoerinligIigiire 2-1 from left to right: Ingo of many mutual concerns, Including

the characterismtios of Intelligence tar-e Various modes of Intelligence col- gets, collection capabilities, use of col-lectIon (SIGINT, IMINT, HUMINT). lectlon capabilities, and the operating

e Single-source correlation and pro- environment. These models aredlul. reflected In training plans, in operatingdoctring, and ?'i knowledge bases usede Multi-source aggregation and by analysts.

fuilon (over time and sources) .ac, ordIng to mission orientation 2.2.2 Formal Information Flow Arrange-(IPB, Targeting, Situation Assess-"tment, OPSEC). Linking the different special work set-e Specific client-oriented Informs- ting are flows of Information throughformal communication channels (recordand message traffic, mail, command and

In the figure, main data-flow paths are control communications networks, writ-am left to right, with a corresponding ten orders, etc.). These carry require-increase In e igregation of data. ments, tasking, raw data, filtered data,2.2 Communication Factors and analytical results and products.Besides the main data flows, Figure 2-1Indicates that the various types of work

- 2-2 -

66144,~. -


25/74

2.2.3 Informal CommunicationsInformal communications (via securetelephone, meetings, ronsultatiois withpeers, Interactive digital corr-miict-tions networks, and personal lei.er.-)link the work settings, proviclrganalysts with clarifications, feedbact,and ready extensions of their ownanalytical capabilities. A functdon oftenserved by informal communications Isthat of making analysts aware of largerperspectives. Interviewed analystsobserved that when they were leftunaware of the "big picture," they feltIso'ated end were less able to advancetheir skills. This condition was cited asbeing more prevaleat In tactical unitswithout an active mission. Analysts Instrategic facilities do not exhibit thisconcern, because of larger local analystpopulations and better facilities forsharing Information with other organiza-tions.2.2.4 Job RotationJob rotation through different work set-tings serves to link the settings.Analysts report that receiving a varietyof tactical and strategic assignmentsdevelops a perspective about availableinformation sources (including personalcontacts) that are used in analysis.Aalyata who heve strategic assipg-ments prior to entering the tacticaienvironment know where to obtain tac-tically useful all-source Information.

-2-3-,i

L.....--


26/74

ADAPTATIONMONITORING AND EVALUIATION

R UI/"PROCESSING. INITELL


27/74

8.1 Resouere Management Analysts must frequently support multi-The resources allocated to Intelligence pie missions concurrently and divideanalysis alldI i their time between several tasks.Analysts may serve several clients9 Time. when producing analytical results andmust make compromise decisions on for-" Analytic personnel capabilities. mat and content. The analyst must also

"eInformation flow. find time to secure the long-term bene-" Information storage capacities. fit of sidll maintenance andknowledge-base development. Annlysts"* Collection assets. who were Interviewed felt that they"* System facilities, had lees control over this last categorythan was necessary to develop and"eResource management capabill- maintain their proficiencles.tis. 8.2 Adatatdon"* Adapth',o napabllitles. Adaptation Is the development of

As Indicated In Figure 8-1, resource appropriate new responses, by what-management tasks Include planning, ever means, In tie face of significantallkA.ation, and tasking. Each of these change. Figure 8-1 shows the adapts-tasks Involves decision making by tion function as a key feature of theanalysts as well as management. Vari- generic Intelligence production model.ables considered In resource manage- Adaptation of the Intelligence produc-ment decision making include: tion process requires that both the

9 Level of threat. erganization and the analysts examinethe quality of products and use what-e Availability of resources. ever means are available to correcto Priorities. deficiencies. Adaptation takes place atboth IndMdual and organl Atonal levels.e Short term vs. long term benefits. 3.2.1 Individual Adaalon

One of the most Important managed It Is Impractical to tightly control intelll-resources Is time. Mechanisms used to gence production by external monitoringdjust production to time pressures adqa~yrlbo hcs h ultInclude:and quality-control checks. The qualityof a particular piece ;of work can be9 Segmenting production Into phases checked only by replicating the entireoriented to product suspense process, uoing another analyst withtimes. better qualifications, and then making acomparison. This cannot be done on aControlling the size of geooraphic large scale because of the high cost ofarea, or number or tn es of tar- the duplicative effort. Thus, the adap-

gets handled by one analyst stie capabilities of the Individual* Assigning priorities so that time- analyst are at a premium. individual

sensitive data gets handled first. adaptive activities are exemplified bythe following behaviors observed duringn the context of producing specific st iissite visits:products, the analyst makes manyresource management decisions.

-8-2-


28/74

.e Making voluntary judgment about e Analysts remigned to the newown performance. area will have a substantial-a Seeking Informal fevidbaok chan- start-up time to become familiarhels.' with the targets and achieve con-fidence In Interpretations..e Using non-standard processing o Sociocultural knowledge may notstrategies (whatever works). 8ooulrlknwegmyntbe current for analysts who must

.e Using own Initiative for Increasing work a new geographic area.awareness in significant areas The administrative adaptation processIs Important In the study of intelligenceSeeking cues from other Inteili- analysis because It contributes factorsgence sources. basic to analytic performance. One ofSeeking cues from associated the strengths of strategic intelligenceorganizations Is that they are able toelements or processes. use highly experienced and qualifiedDeveloping contacts In Intelli- analysts In the production process.gence community or within These analysts accumulate a largefacility, store of knowledge on specific target

areas and are able to perform in-depthto aid In distinguishing cultural Interpretations. SociocuIturalknowledge at the strategic level can bebackground features from s1g- supplemented by immediate access tonificant data. State Department and civilian sources.Building continuously theknBuledinge ontiuouy thes Because strategic production resourcesknowledge of "how things are limited, some target areas'must bework" to aid In interpreting assigned lower prlorities. However,6~parae data (e.g., maximum asindlwrpoite.Hev,speed and Incline capacities Of computer data bases maintain archivalsehiled, an iecapo ractes, files of target Information stored earliervehicles, weapon ranges, that can be recovered easily after longoptimum transmitter positioningrequirements, etc.) periods of lowered attention to particu-lar targets. Such data serve as base-

3.2.2 Admlnltratlve Adaptation lines for renewed closer attention.Uk. the Individual analyst, the Intelli- Comutilcation networks make suchgence organization utilizes various data available on a shared basisthroughoutntzetstrategiczcommunity.adaptive mechanisms when faced with throughout the strategic community.new problems. If a target area that On the other hand, Intelligence produc-was not being followed In depth sud- tion facilities in tactical echelons belowdenly becomes Important, the organiza- corps are faced with adaptationtion may face several adaptation prob- requirements every time the supportedlems: tactical unit must prepare for a con-

e Inattention to the area may have tingency operation. The tactical Intelll.,resulted In gaps In reference gence unit Is at a disadvantage inmaterialsn adapting to new contingencies for anumber of reasons. Analysts are likelyto be less experienced than at the

- 3-3 -


29/74

IN I NNWATM PIRMORIAcx cMLATIVEY SPARSE KNOWLEDGE PROM INPOMTIAT WITERPRETIE

DATA PATTERN INTERNAL AND EXTERNAL PILTERING PRODUCTSAS NIPUT MOERY AS INPUT DECIIONS AS OUTPUT

-, MESSAGES-, = REPORTS) MITERPRTE SUMMARIES_-) PROCESSES ESTIMATESANNOTATED DATA

INFORMATION iNFORMATION OUTPUT REDUCEDCONTRIBUTION CONTRInUTION CONTENT INFORMATIONLOWEST MQHEST CONTROL OUTPUT

WOM"ATION CONTRIUTION TO INTERRETATION RESULTS

Figure 8-2. Sequence for Intelligence interpretation Ta3kstrategic level, and highly experienced short supply. Real data may be unavail-analysts may be unavailable to assume able for analysts to work with. Interac-training roles. Persinnel shortages may tion with community intelligencedraw analysts away from analytic traln- resources Is sometimes difficult.Ing duties. If the unit does not have a In the worst case, the tactical unit mustcurrent active mission, analysts may not depend almost entirely upon the mdlvi-e current on the Intelligence threat dendaosetieyun h idi-andmaynotbe gurantroreisontedInea atin dual skills and backgrounds of analystsand may not be garrisoned in a ocatonrequIre-close to the resources needed to main- ts.ttan knowledge for areas of likelyoperation. Storage facilities are often 3.3 Interpretationnot available for archival data thatwould be useful In preparing for a wide Although the generic production modelrang ofntigences, ersonel depicts Interpretation as only one ofrange of contingenciss. Personnel three main analytic functions, It Is, Inclearances and billets for sensitive fact, the basic one. Figure 3-2 shows

the Interpretive processes as a cycle-,-I


30/74

"INTERPRETIVEmirupte

Figure 3-3. Interpretive Complexity as a Functi~on of Data Richnessand Number ofUpcrtable Interpretaftiosof (1) receiving Incomplete or sparse process.data, (2) Interpreting the data, and (3) 3. mrr~eCmlxt

331InterpretivaCmlexit

formulating the Intrllhgence product(mFssages, estimates, etc.). The oen- Interpretive on laxity Ihne omposisetral function of Interpretation Involves evaluatio of the relative difficutty ofmaking many Information filtering decl- anblytec tasks. Interpretive complexityalone In the pro(1) of cverting I Inverely proportional to the rschnesssparse or uneven Input data to Intellr- of the Input data and d(rectly proper-genoa pra ducts. The basis for such tional to the number of reasonabledecisions s usually memory-stored con- alteramtives against which an Intorpre-ceptual models of terpeti of real- tive devlsion must be made. Figure 3-3world objects or events thought to Illustrate this relationIne p. Thus litua-have generated the data being Inter- tInvswhich are the most difficult andprsted u complex to Interpretare those In whichthere are few data available but forFigure 3-2 also summarizes our chser- which many viable interpretations arevatcon that, of the Information going aliworldnteobrectsve prordeventsa htto possible; other situations In which therede manysdata ar t point to a singleamount co~mes from conceptual -models obvious Interpretation are low In inter-stored In memory (both Internal and pretive complexity.external memory), than from the databeing Interpreted. In other wordet, Intel- Another factor In Interpretive complex-Ilgence analysis is usually more a Ity is the level to which the Input dataconcept-driven than a data-driven have been aggregated. Data may be

...- 8- .


31/74

--.

aggregated over periods of tin.e, over The accurate and timely identificationmultiple sources, or In levels of of three requires the use of muchabstraction (e.g., n artillery pieces vs. a sophisticated knowledge organizedbattery). Aggregation of input data can around conceptual models. Potentialwork positively or negatively in the threat is Inferred at the national levelInterpretation process. If the aggrega- from scenarios (conceptuq! models)tion Is done properly, the richness of depicting capabilities of:the input data is Increased and inter- a Military forces.pretive complexity is lessened. On theother hand, aggregation can Interject * Weapons.biases due to normalization, to Incorrect s Strategic Industries.interpretation fi abstraction, or to remo-val of valid ambiguities in preprocess- 9 Economic factors.Ing. e Organization and distribution ofBased on oue observations three con- power.cluslons about interpretive complexityhave been drawn: e Strategic and tactical dis" sitlcnof forces and weapons.

e The more complex interpretationtasks are usually performed by e Electruiic warfare capabilities.more experienced analysts better Interring Intentions to use such capabil-able to cope with higher levels of Ities Is more difficult. Conceptualdata aggregation. models of necessary buildup patterns

a for threatening behavior are used to* Interpretive complexity is often Infer current and future Intentions. Thegreater In work settings using high models Identify categories of indica-levels of data aggregation, tecuseanlyss mstmak del- tionw and warning signs that arebecause analysts must make deci- observable thrnugh collection assets.sions on the Interpretation quality Examples of such Indicators are:of their Inputs as well as maketheir own aggregation choices and 9 Provocative diplomacy.Interpretive decisions. a Threats/ultimatums.

9 There Is sometimes distrust of.Interpretations received from work * Hostile political actions.settings at 4ower levels of aggre- e Intimidation.gation unless the using analyst e M military action.has trust in the source.

2 Major military action.3.3.2 Knowledge used In Interpreta-tion e Military augmentation.The requirements for the knowledge a Domestic confilb-t.needed to perform interpretations havetheir origln In one common definition of * Exercise.Intelligence: * Peaceful diplomacy.Intelligence Is the Identification of e Verbal conflict.threat; the two oompoijents of threatare capabfltlWs and Intentions.I - 8-6 -


32/74

In a parallel process, tactical threat Is Models of enemy surveillanceinferred from scenarios depicting capa- capabilities and effects of terrainolitles of such things as: and ECM.9 Poeitioning and capabilities of Indlvi- e Political organizations and control

dual large weapons. of power.* Positioning and capabilities of mill- e Command and control organiza-

tary units of significant size. tions.* Movements of weapons and units. 9 Personality prgfily.e Positioning and movement of logisti- 9 Terrain models for use in traffica-

cal units and headquarters. bility or tactical dispositionThe best-known threat model is Order analysis.of Battle (O). The O can be imple- Ethnographic models.mented In hardcopy or computer database form. 05 provides capability e Chain of events.models of force structures, weapons, 9 Time and frequency patterns.status, and disposition. OB also pro-vides Insight Into intentions of the e industrial production models.forces by implication from training, Weapon system performancereadiness status, tactics, personality of models.commanders, and strategic or tacticaldisposition of the forces. OB data Is e Economic models.used at alh echelon levels, but Is e Unes of communications.tailored to the 0eo",raphic area ofresponsibility and to a level of detail e Network models.that Is meaningful to the current situs- These models may be used to transferriga. general characteristics of knownIndicators of threat guide the develop- threats to unknown threats by analogy,ment of further Intelligence Information and to develop higher levels of threatfor determining vulnerabilities and plan- Information by piecing together Incom-or etem~nnglnrablltls ad pan- plete fragments of data (such as In thening responses. In planning responses, Cuban missile crisis, when thebehavior modeling is used to develop existence of Soviet missiles in Cubaand to authenticate models of possible wastasce fromit shaes ofantagonist reactions. These models was ascertained from the shapes ofpredct robbleantaonit rsposes shipboard containers and from tracks ofpredict probable antagonist responses hayvhce rudsmlrsaeunder given sets of circumstances. The heavy vehicles around similar shapesseen Inland).importance of behavioral modeling lies Inits ability to project "what if?" situs- 3.3.3 Using InterpretiveModelstions. Because of the predictive nature The main processes of interpretationof these models, they produce specula- involve the extensive use of concep-tive, probabilistic results and require tnalme the sills ta enaethcontinuous refinement and adjustment tual models. The skills that enable theto maintain their usefulness. analyst to exploit conceptual modelsInclude:Other types of models frequently usedIn Intelligence analysis are:

"- 3-7 -

.,F . r


33/74

Formulation of conceptual models. characteristic).,2e Storage and retrieval of data Interpretation or ambiguities by

using models as the indexing matching to multiple t odels (suchframework (such as situation as a unit that could belong tooverlays, 09, Installation files), either a motorized rifle regiment ora tawk regiment). Communication with clients by uaaof models to provide extensive * Recognition of uncertainty inInformation to supplement the con- interpretations by determining par-tent of messages (such as clients tial matching of events to modelswho have local OB files, mission (such as being able to locate andplans, IPB overlays, etc.). Identify only 80% of the maneuverunits In an observable forcee Correlation of observable real whose structural model calls forworld events with model parame- 40% more).ters to deduce capabilities orIntentions (suah as matching unit 3.3.4 Filteringof Productsmaneuvers to tie beginning of a In addition to the conceptual model han-mobilization). dling skills, the intelligence analyst must

a Use of models to Inductively have the Important decision-makingdetermine the threat from unknown capability of controlling and filteringelements where analogies exist to data output to the client. In decidingk,-own threats (such as extrapola- which Information should be passed ontion of Soviet military doctrine to to the client, the analyst uses at leastsatellite countries), these three mechanisms:

a Inference of event significance by & Determining change from previouscarrying cause and effect rela- products.tionshps to logical ends (such astheImpacts of weapon augmentation), client and utilizing informal feed-

e Fusion of data by abstracting to back to adjust the content of pro-higher levels of behavior (such as ducts or provide clarifications.lower echelon units Into higher e S the output productechelon units or Individual targets form such that it Is self-defining inInto target clusters), terms of Its relation to clients'

e Compensation for Incomplete Intel- areas of Interest and quality ofligence data by fill-in from models Interpretation.(such as using unit Identification The control of Intelligence productand location to identify probable dissemination Is largely a manual pro-units). cess Involving large distribution' lists

9 Attention to serendipitous events that evolve over an extended period ofthat do not correlate with known time. In some work settings, an analystmodels (such as detecting a new must designate action and Informationweapon system that Is dis- recipients, select a product form,tinguished from an old system only assign precedence, and select aIn one externally observable transmission mode. These decisions

- a-8-I1


34/74

determine the content and speed ofdelivery of the productIntelligence products that are highlyformatted and transmitted electricallymay be processed directly Into clientcomputer data. bases. When the clientdoes niot have computer assistance, theproduct flow rate must be controlled bythe Intelligence production source toprevent saturation of the routing sys-tem and the recipients.3.3.6 Summary of Intelligence Produc-tion Model CharaoterlatlcsTo summarize, the generic model ofIntelligence production processes Iden-tites three main functional areas:adaptation; resource management; andInterpretation. The general types oftasks, such as monitoring and evalua-tion, production planning, tasking, pro-cessing, analysis, reporting, and specialproduct design serve these three mainfunctional areas.The central function of InterpretationInvolves making many Information filter-Ing decisions In the process of convert-Ing sparse or uneven Input data toIntelligence products. The basis forsuch decisions is usually memory-storedconceptual models of the types ofreal-world objects or events thought tohave generated the data being Inter-preted.Comparison of the Information consti-tuents of Intelligence products with theInformation constituents of the datasupporting the products shows thatoften a large proportion of the informa-tion In the products has been addedfrom memory. Thus, the Interpretive pro-cess Is often more concept-driven thandata-driven.


35/74

4. ENVIRONMENTAL VARIABLES minlm'zed to make efficient use of rela-Environmental or work setting variables tively scarce analytic personnelresources. Nevertheless, requirementsffecting analytic performancA are veycmo.Teitlg.corn-Important In the development of te for maintaining overlapping knowledgeScognitive model because they cir- for different areas of responsibility areSo~~~~umscr ibe the range of perceptions, eyc mo.T ntlgaecou-' deisin mkin conideatinsand Ity attempts to develop and exploitmnakin conderatina s, aed- such overlap through working groups,which rehe a nalystxpert information distribution, Integrated com-esiputer data bases, and Informal communi-tasks In a given work setting. Thescope of environmental variables cations.affecting the cognitive processes of 4.8 Client Orientationsanalysts is summarized in the diagram In Figure 4-1 shows client orientationsFigure 4-1. affecting the analyst. Attention to a4.1 Assigned Domain particular client can Improve the valueof the analysis product for that client.geograssigned udomarinusu cnistsn oh This goal Is sometimes difficult for thegeographic boundaries within whichto achieve In practice, becausedata representing certain target types an y ts are in eously aseare to be Interpreted. Deep familiarity by c tion procssuand tero-Swith geographic, climatic, and sociocul- by the production process and the pro-wt georapic clmaic anus reqir inluiv otural aspects of the assigned geo-graphic domain Is a basic asset to the compromise format The importance ofIntelligence analyst. Intimate familiarity one client's needs over another alsowith the domain aids In identifying mill- changes.tarily significant targets and in anticl- Tactical Intelligence production unitspating constraints on military opera- have multiple client orientations both intions. The domain variables depicted as garrison and in deployment. TacticalImpinging on the analyst In Figure 4-1 Intelligence units are typically attachedbecome especially obvious during a to a tactical unit commander, andchange of geographic domain. When- thereby assignec main client. At theever an analyst Is reassigned to a dif- same time, the ta Jcal intelligence unitferent geographical domain, even also has the general intelligence corn-though the types of targets may be the munity as a client. It Is Important tosame the analyst usually requires a note that the "perceived" client Is moresubstantial start-up time to regain peak critical than the actual end users of aperformance levels. The same Is true of product; an Individual analyst or anreassignment of target types within the intelligence unit may adapt to satisfysame geographic boundaries, some goal, but thel Intended user is not4.2 Management and Tasking always the sole beneficiary.4.24. ManaememndpportsnManagement directives define the 4 System Supportsdomains of an analyst's geographical The supports available In the intelll-and functionav areas of responsibility in gence production facility are importantthe overall production picture. These to analytic performance. Computationaltasking constraints change with time. aids and secure communications areOverlaps in assigned responsibilities are valuable assets. Storage space for

- 4-1 -

|


36/74

DATA TYPES DOOMA MGM &TASIUNG490GAPfIC ARMA OGMNE1"NOGRAPI4IC

wumERVEns TWRA"N FEATURES AsEamO NTRSPETERE DATA SOCIOCULTURAL AREAOFC ACCOUNT.REMILTS ECO nMi C TASMUN AGENCIESEXPECTATIONS WEAThER WORK UREAWDOWNuOSE~.8CHECK, EVAL, CORR

CLIENT ORIENTATIONISEPO AQUIS~iONTACTICAL COMMANDERCNOLLECaUTION plWPREDICTABILITY ANALYST 10111 pOVERLAPS FIRESPLAN&LOGISTICS.OThER INTL

JU04METAL SYSTEM SUPPORTSPRIORITIES KNOWLEDGE BASE DISPLAYSTIM AVAILABLE DW aNTR DATA BASESCOMMNICAIONCURENTCOLLEAGUESCMMUNICATINBW REFERENCESSISIUPCMICEANTICIPATED

EXTERNAL COMM.

Figure 4-1. Variables In Analytic Work Settings

maps and secure data can make a sub- Intelligence environment Is maintained Instantial -difference In analytic perfor- the form of materials such as refer-mance. Film processing facilities are ences, archival data files, an d computerImportant to IMINT In the same way that data base systems. Computer aids areelectronic signal recording and repro- thought -to be essential for extendingcessing equipment Is Important to SIG- the knowledge base to Include archivalINT. data that cannot be effectively handled4.5 Knowledge Bass aulyThe knowledge base or externai store'of Information applicable within a given-42


37/74

4.6 Judgmental Criteria airorew after the mission.Analysts must exercise judgments on a 4.8 Data Types Familiarity/skillsfairly large number of dimensions. Many analysts are cpeciized in oneThese concern the possible signifl- particlarydtame specialize -canoes of new data for the needs of particular dati mode. This specialize-various analytic clients, the moment- tion In formalized by military occupa-to-mmenpimu useof he naltic tiona[ specialities structured along theto-moment optimum use of the analytic lines of data capture modes (voiceresources actually available to each intercept, manuel morse, imagery, etc.).analyst, and consideration of the infor- Specialties are also defined formation loads Imposed on various ana-lytic clients. Furtheimore, these Judg- analysts who handle analytical resultsmental factors are not static: needs of rather than raw data (cryptanalyst, OB,ll-source).clients often change rapidly; the load ofIncoming data to be analyzed changes The type of data being analyzedcontinuously, the analytic resources Impacts the analytic processes in aactually available to the analyst often number of ways. The related mentalchange, and the information loads on imagery differs for various data modes.various analytic clients also varies. As The Information properties (comparativea result of one or more such considera- limitations, strengths, "artifactual" con-tions, the analyst must often modify tributions) of different data modestask priorities and the time allocated to reporting the "same" phenomena alsoanalytic tasks. In a crisis, Incoming vary markedly. The difficulty levels andIntelligence data is usually voluminous, amount of experience required forand client saturation may occur If the acceptable performances vary for dif-analyst does not actively filter analytic ferent data modes used to report vari-results by significance and client ous classes of phenomena. Theneeds. This Is particularly Important, for analyst's "mental world vlewpoirnt" isexample, when the client Is a near- conditioned by the particular data modereal-time Intelligence fusion certer. being used.4.7 Information Acquisition Very little cross-mode training existsThe analyst normally has no direct until the analyst moves to a manage-Thepanalystncorllyio haisinon. d t ment role. Within Imagery Interpreta-Impact on collection decisions. How- tion, there are many submodes (IR,ever, the analyst usually has sufficient SLAR, forward oblique, overheadknowledge of the collection plan to stereo/mono, video) that annalyst willknow what type and how much data will htero/moncy withat a cn a -be podueda isson.For xamle, have proficiency with after t, combine-be produced by aission. For example, tion of tactical and strategic assign-n a tactical reconnaissance mission, ments. Substantial ekill-develonmentthe original threat analysis may have menIs Substanal s pecificdetermined the critical areac for sur-veilllnce and projected the ty-pes of data-capture mode In al l specialties,targets; terrain, and concealment to be and particularly In SIGINT.found In the area. The analyst maytherefore be Involved In planning theflight path and briefing the aircrew onair defense threats. Additional Informs-tion may be gained by debriefing the

,4.-~..-.... -... . . . . . . .


38/74

6. ANALYST VARIABLES 6.1.2 Science arnd TechnologyA majo" conclusion from the work per- Most specialities of intelligenceformed on this project Is that great analysis use almost all for, s of physicalImportance must be placed on the attri- science and technology. A strongbutes of the analysts tasked with pro- technical background helps the analystduclng Intelligence. The main attributes understand what Is seen br heard Inneeded are In the areas of: data collected from technically complexe Relevant knowledge, targets such as alriraft, missiles,"radars, communication nets, factory"e Skills (perceptual; conceptual; complexes, and the like. The complex

communication), workings of weapons systems, commun-" Orientation toward managing Ications, command and control organiza-relevantons and procedures, and electronicssystems must be understood for proper"e Personality. Interpretation of Intelligence data In

5.1 Knowledge both strategic and tactical missions.The knowledge relevant for analysis Emphatically, this principle holds for allIncludes: levels of intelligence analysis asdefined earlier. Although the Inexperi-* Military science. enced Image Interpreter performing a

* Science and technology, target location or counting functionrequires much less general knowledge* Sociocultural knowledge. and understanding than an analyst withe Sensor mode knowledge. a more explicit evaluation role, all levelsbenefit by a thorough understanding inBackground In each of these areas aids their area of responsibility.

the analyst In performing most analyti-cal tasks. Commonaiities and differ- a.1.3 SocioculturalKnowledgeences In knowledge requirements for Knowledge of a target country's cultutedifferent analytic specialities are Is valuable In discriminating militarilyhighlighted below, significant items from non-significant5.1.1 Military Science ones. Ethnography helps the analyst

know where to look for significant ItemsCommonalitles between SIGINT. HUMINT, and what to filter out. SocloculturalIMINT, and FUSION speclaitleb Include knowledge Is particularly Important inan extensive knowledge of military sci- strategic Intelligence analysis whereence areas Including Order of Battle the use of economic, political, andand military geography. Characternati- Industrial models may be Involved.cally, this knowledge Is acquired inspecific relation to the targets and 6.1.4 Sensor Mode Knowledgegeographic area within the analyst's Because the entire Intelligence produc-assigned area of responsibility. The tion cycle is included In the scope ofanalyst gradually expands this intelligence analysis activities, the ablil-knowledge base by cross-training or Ity to understand the capabilities .ofreassignmet to other Intelligence pro- sensor systems Is a fundamentalductkon facilities, knowledge requirement. Although an

- c-1 -

iL _-


39/74

Imagery Interpreter may not need to 5.2.2 ConceptualSkillsknow how a particular camera works, The generic P telllgence productionthe constraints under which photo- model presentad earlier depicts In.ailli-graphic data Is collected must be genoa analysis as predominantlyunderstood when making decoisioaabout collection plans and the quality o concept-driven rather than data-driven.altllection pierre nd quThis conclusion was reached afterIntellgence products, observing the extensive memory6.2 Skills resources (see Figure 3-2) thatSkills Involve knowledge, but are dis- analysts bring to the production of mostIntelligence products, and noting thattinguished here from knowledge In terms muhotemmoybsd i fma i iof added roqulrements for highly prac- muconptui nature.bThe proaui-as conceptual In nature. The produc-tioed performances. tion process Is also concept-driven in6.2.1 PerceptualSills the sense that new intelligence pro-ducts ca n be, and often are, producedThe phrase perceptual skills, as used In the absence of new data. Inteili-here, means practiced performances Indiscriminating between things on the gene preparation of the battlefield(IPB) is an example of an activity thatbaal ofsubleffrenes n hw te~t depends almost entirely on previouslysound or look. Perceptual skills are developed and stored knowledge. Asespecially Important for single-source Indicted earlier In this report, highlyIndictgenc eradlieonher thi reothglIntelligence production. There Is a experienced and effective analystsstrong consensus in the field that the often appear to organize the mentalperceptual skills fundamental to SIGINT storag e f owg e aon contand IMINT single-source production strgofuhknwee ondc-andt bMe develousingi u rceaduct in ceptual models. The analyst's ability to.mTht isthe refinoped perceptureal skis deal with conceptual models grows as aThat Is, the refined perceptual skills result of experience In learningnecessary to Interpret particular kinds abstract concepts. This type of learn-of Images or to recognize certain voice Ing appeara t3 be more In lina with thator certain morse or telegraphic ypera- received In higher educ&tion than Intore can'not be fully developed in exer- trade skill teaming. Characteristically,clses disassociated from real targets. analysts who progress to higher skillThe analyst's confidence and speed in levels have had post-high school edu-Interpretation and In the ability to catlort.detect significant changes also depend 6.2.3 ComaunicationSkillsheavily on familiarity with the particulartarget area. A substantial time period Foreign language Is of special Impor-Is required to gain such familiarity even tance to COMINT and HUMINT Intelil-for experienced analysts. SIGINT gence production for obvious reasons.analysts stated that It would take two From a general 3tandpoint, languageto six months to adapt to a new target skills are vital to xhe analyst for under-area after being reassigned. IMINT standing the meaning of collected Intel-analysts are often semi-permanently ligence even after It Is hiterpreted.assigned by geographic area and/or Writing and speaking skills are essentialtarget types In order to take advantage for the analyst to communicate analyticof the resulting accumulation of percep- results to clients.tual familiarity.

- 6-2 -

. .. .. . .. .. .. . . . .. . .. . .. ... .. ... . . .. ...... ," .;


40/74

.x ~mITERNAL MEMORY EXTERNAL MEMORY UPATE/EXTENSIOU

Knowledge of evant. situationKnowledge of procedureKnowledge of Judgorinal/ 4-analyticl criterionKnosa 1ge of how to get(more) Information. Knowledga of who to aM

or om l colleagues- Knowledgeof whore to findInformation ons

OeVO. astuation. Atforsenoe. aids.judgntal/mIsn~ioal historical Wiesoriterionprooedurw Procedures. wamng Ing.

Knowledge of how to acquireInformation which Is not - Crollection reeourc~mamobelb In eWanai memOry

Figure 5-1. The Analyst's Use of Memory Resources

8.3 Memory Management Orientation from storage facilities where they areavallabde to storage facilities whereTh e two of Information from memory Is a the ar o. The oewyarw nicentrally important factor In Intelligence the mrnn e one-way arrows ndi-

analysis. This Includes not only the use cats the manner In which use of thes ther two sources of information Isof-the analyst's own memory, but also dependent upon knowledge contained inthe use of external memory resourcesavoltable In the extended work setting. internal memory.The three columns of Figure 5-1, to be 5.3.1 Knowledge of Events and Situa-discussed in mae detail below, depict tionstypes of Information contents available Figure 5-1 shows that the analystfrom Internal and external memory, and retains an Internal memory store offrom the functions for Information on situation and curentupdating/extending external ,kibmory. In ts. An Imagery IndreurrThe twc-way arrows Indicate the evsnts. An imagery Interpreter, forreciprocal flows of types of Information Instance, may become so familiar with a

- 6-8 -

L-5-3-YN.


41/74

terrain area as to be able to detect any questons of colleagues when uncertain,type of change that may be slgn' ,,ant. since this process continues to be part8.adlarly, a SIGINT analyst may be able of many analytic problem-solving situa-vo recognize targets on the basis of tions. The shared group value for "ego-"personality" characteristics that are less" problem-solving exchanges wasnot-definable in terms ot recorded sig- frequently observed.nature Information or signal content. Anana~ystmayais revew xtenal 6.3.3 Knowledge of Accepted Judg-analyst may also review external m na n nltclCieimemory resources, such aa situationmaps, status displays, and message The priorties and significance oftraffic files, to provide cues that, analytical results -,re establishedthrough mental association, may lead to through cooperation between the intelli-extending an area of search or expand- gence organization and the client. AtIng the possible interpretations, the strategic level many of these cri-

teria are long-lived. In tactical environ-ments, the analyst Is frequentlyFigure 6 shows that knowledge of pro- required to make judgments of changingcedure resides In internal memory. significances and can therefore profitOther than manuals for operating partic- from frequently renewed contacts'withular types of equipment or for format- clients, In particular the unit com-ting products, there Is almost no docu- mander.mentetion on the detailed procedures 6.3A Knowledge of How to Get Moreused In performing analysis. Such pro- .n Kormotioncedures are conveyed to the analyst Informationthrough on-the-job training. The neo- The analyst normally works as part of aphyte analyst geoierally gains profi- group of co-located analysts who shareciency by working closely with experi- the Intelligence production workload andenced analysts who provide assistance operate as a team-memory resource forand Immediate feedback on perfor- one another. As Indicated by Figure 5-mance and answer questions on tech- 1, colleagues are the most Immediateniques and Interpretations. This face- source of knowledge after the internalto-face learning relationship is very memory of the analyst Is exhausted. Inwidespread and recognized as essential the operation of several intelligenceIn intelligence analysis. It Is referred to fusion groups during tactical exercises,by some experienced analysts as the It was observed that specific areas of"oral tradition" of learning inte!l!gence memory responsibility were designatedanalysis. to Individual members of the analyticalThe oral tradition fo r learning ths pro- team. An analyst assigned a specificcedural Intricacies of nti l l lgonce content area was responsible for col-cedual ntriacisfintllignce lecting, reviewing, and remery~beriki~g ornalysis emphaslzeb the Importance lelnrvwig drmebrigoanalysiforma ze-th-fce imporance- maintaining availability of all Informatioc,tono l apertinent to "hat area, as well as briaf-Ing the fusion tesm chief when neces-In this regard, it Is Important that the sary.analyst develop confidence In the abil- Association with expelenced analystsIty to analyze and at the same time Is the major source for gainingremain comfortable about asking knowledge of where to find Information,

a


42/74

Analysts develop contacts from past can be ierformed more effectively withassignments or training courses and somewhat closer coordination betweenretain Informal communication with those the two functions. As a result, It Iscontacts. Mlssion-defined units of sometimes fLiund advantageous In theIntelligence organizations are not large, tactical environment for the intelligenceand analysts with wide experience are analyst to be involved In developmentwell known In both tactical and stra- of the collection plan at the earlytegic settings. stages of deployment, and to providecontinuous feedback on the quality ofMost analysts supplement personal collected data. For example, a SIGINTmemory wit

Documents

Human Processes in Intelligence Analysis: Phase 1 Overview