TRAINING CRITICAL THINKING TO ENHANCE

BATTLEFIELD INITIATIVE

Marvin S. Cohen & Leonard Adelman

Cognitive Technologies, Inc.
4200 Lorcom Lane
Arlington VA 22207
703-524-4331
mcohen@[Link]

ABSTRACT

The knowledge and thinking strategies associated with taking initiative were investi-
gated. Training based on the results proved successful in improving tactical battlefield
decision making performance.

INTRODUCTION way to accomplish the higher-level purpose of sup-

porting the division in a timely manner?
In a rapidly changing and uncertain task envi- How can individuals and teams be trained to
ronment, there is sometimes a tradeoff between act- make decisions that balance the benefits of initiative
ing quickly based on familiar responses to past against its risks? Our approach does not start with
situations and taking time to find a unique solution an priori normative model of decision making, but
to the current problem (Cohen, Freeman, & Wolf, with empirically based prescription. We first ask,
1996). At the organizational level, a similar tradeoff what strategies and knowledge do experienced deci-
occurs between risky exploration of new markets, sion makers draw on to make decisions about the
technologies, or tactics versus falling back on exist- appropriate degree of initiative in unexpected situa-
ing strengths (March, 1996). Sometimes, individual tions, and how (if at all) do less experienced deci-
and organizational dilemmas intersect: When time sion makers differ? This question must be answered
is constrained and communication is difficult, indi- by a combination of experimental studies, analysis,
viduals acting within an organization may have to and cognitive modeling. We then develop training
choose between coordination with other team ele- that attempts to convey the relevant strategies and
ments and taking initiative to seize fleeting oppor- knowledge to less experienced decision makers. Fi-
tunities (Cohen & Thompson, 2001). This is a not nally, we test the training to find out whether per-
uncommon predicament in combat: Company A’s formance of less experienced decision makers is
job is to guard Company B’s flank while Company improved. In this article, we will briefly describe a
B secures a bridge that the division intends to cross. case study of this method. We analyzed critical in-
Now, however, Company B appears to be stalled in cident interviews with experienced Army officers,
a major firefight some distance from the bridge. modeled the results, developed computer-based
Company A cannot raise either Company B or training, and tested it with students at the Army
higher headquarters on the radio (and it will take Command and General Staff College (Cohen,
too long for runners to find them and return). Thompson, Adelman, Bresnick, Shastri, & Riedel,
Should Company A sit tight until Company B is 2000).
ready to seize the bridge or until communications
are reestablished? Should it go help Company B in ANALYSIS OF COGNITIVE SKILLS
the firefight, at the risk of getting bogged down it-
self? Or should Company A attempt to seize the Initiative means taking “the first step, or the
bridge now – a risky choice, but possibly the only lead; the act of setting a process or chain of events
in motion” (Brown, 1993). Interventions are often organization (i.e., above the level of one’s own
(though not always) targeted at the decision-action- unit). Moreover, hierarchical analysis of correla-
outcome cycle of other agents. In business, for ex- tions (Johnson, 1967) revealed three basic clusters
ample, one may try to influence, predict, or react to of mental models, corresponding to the three time
the actions of competitors, customers, superiors, orientations: reactive, predictive, and proactive.
subordinates, or co-workers. In combat, one may try Proactive decision makers, who tried to impose
to influence, predict, or react to actions of the en- their will on others, were most likely to think deeply
emy, other friendly forces, superiors, or subordi- about their own higher-level purpose. Mental mod-
nates. In all these cases, the earlier and more influ- els of the other agent’s intent and of one’s own in-
ential the intervention, the more initiative the person tent (at the level of one’s own unit) were utilized
has shown with respect to the targeted process. The approximately equally often by decision makers
essential questions for training are: What are the who were trying to predict intent as by decisions
mental models and the thinking processes that un- makers who were trying to influence it. Predictive
derlie initiative? decision makers were more likely to use a rate of
We analyzed 25 critical incident interviews and movement mental model, while mental models of
problem-solving sessions with active duty Army reliability of evidence were used both in the predic-
officers serving on operations, planning, and intelli- tive orientation (to evaluate predictions) and in the
gence staffs from the battalion to the corps level. reactive orientation (to figure out why a prediction
We identified clusters of topics that tended to be failed). Alternative causes and effects were consid-
correlated with one another within and across inci- ered most often in reactive modes, when decision
dents. We will refer to these correlated groups of makers tried to explain a failed expectation. Mental
concepts, together with their implicit or explicit models of action sequence, which specify how
causal relationships, as mental models. Among one’s own actions are to be carried out, were ap-
these were mental models representing three de- proximately equally important in all three time ori-
grees of initiative, or time orientations, defined in entations.
terms of where and how they intervened in the In sum, concepts in this domain appear to be
chain of events representing another agent’s deci- organized into a set of mental models, including
sions, actions, and outcomes. The proactive time purpose, intent, action sequence, alternative causes
orientation was present if a friendly action was de- and effects, and reliability of evidence. These mod-
signed to influence future enemy or friendly intent. els in turn are organized around a set of more fun-
The predictive time orientation was present if a damental principles pertaining to initiative, or the
friendly action was adopted because a future enemy time and manner in which uncertainty about other
or friendly action was expected to occur (without agents is reduced. Focus on high-level purpose and
the friendly unit’s doing anything special to bring it proactive time orientation are strongly correlated
about). The reactive time orientation occurred when with one another and with experience.
a friendly action was adopted because of an enemy
or friendly action already accomplished or under- A TRAINING STRATEGY
way. The three time orientations are not mutually
exclusive. A decision maker might be reactive at The training to be described is part of a larger
one level but proactive and/or predictive at other package focused more generally on critical thinking
levels, with respect to other decision cycles that be- skills in tactical decision making (Cohen, Thomp-
long to the same or different agents. Finally, we son, Adelman, Bresnick, Shastri, & Riedel, 2000;
modeled the relations among time orientations, Cohen, Freeman, & Wolf, 1996). The part of inter-
other mental models, and degree of experience in a est to us here teaches students the elements of initia-
multidimensional space. tive, focusing on how to think critically about pur-
Years of experience was significantly corre- pose and time under conditions of uncertainty. Each
lated with use of a proactive time orientation as well segment of the training contains an introduction to
as with reference to higher-level purposes of the the relevant concepts, military examples, historical
case studies, and interactive exercises with feed- to increase the chance that predictive assumptions
back. All exercises involve brief, realistic military will turn out to be true, while reactive tactics moni-
scenarios adapted with permission from the tactical tor for the unexpected. Then, the enemy does in fact
decision games published monthly in the Marine behave in a surprising manner (heading in a differ-
Corps Gazette. ent direction than expected). The initial time orien-
The section on time orientation describes ques- tation now is reactive, to mitigate any immediate
tions that need to be asked to fill gaps in proactive, threat from the enemy action. The next phase is
predictive, and reactive mental models. To create a predictive, to figure out how to exploit any enemy
proactive mental plan, for example, the decision weaknesses that the action exposes or creates (e.g.,
maker asks: “What are my purposes? What do I failing to cross the river leaves a command post
want the enemy to do that will promote those pur- relatively undefended on the other side). At the
poses? And what can I do to get him to do it?” To same time, these opportunities may be used to cre-
create a predictive plan, the decision maker asks: ate new weaknesses, i.e., to proactively degrade the
“What will the enemy do and what strengths or enemy’s capability to pursue future operations (e.g.,
weaknesses are associated with those actions? What by destroying a command post or attacking logis-
are the implications of those strengths and weak- tics). The result is a template for reaction to surprise
nesses for my purposes? And what can I do to avoid that shifts as rapidly as possible from reactive to
the strengths or exploit the weaknesses?” A section predictive to proactive orientations.
of the training on purpose emphasizes the impor-
tance of considering not only the purposes of the FIELD TEST OF THE TRAINING STRATEGY
officer’s own unit, but how it is intended to support
and provide backup for higher-level purposes of Method
superior units, as well as the tasks of adjacent units.
A section on critical thinking about initiative Seventy-eight students at the Army Command
introduces students to an iterative process in which and General Staff College, Fort Leavenworth, Kan-
possible problems are identified and corrected. The sas, participated in an informal study. All students
primary emphasis is on how each time orientation had the rank of Major, and were enrolled in the
can be used to address weaknesses in the other time spring quarter of an advanced tactics course, taught
orientations as plans are gradually elaborated and by LTC William Hadfield. The participants’ aver-
improved. For example, suppose planning begins age length of Army service was 12.7 years, which
with a predictive model based on the expectation included an average of 10 major exercises and an
that an enemy unit will cross a river and be vulner- average of approximately two years of command
able to attack as it crosses. A devil’s advocate strat- staff experience. Over half of the participants (66%)
egy is then used to critique the plan. (An infallible had experience in a combat or peacekeeping mis-
crystal ball says, “The plan will fail. Explain how.”) sion.
This process brings to light hidden assumptions Training consisted of a CD-ROM handed out to
about enemy intent upon which the plan depends. students, who were asked to review it at home. Par-
To make the plan more robust, proactive tactics are ticipants were asked to respond to a different tacti-
developed to lure the enemy across the river. Other cal scenario before and after the training. For this
proactive tactics are developed to increase the en- purpose, two brief tactical scenarios (Sanna’s Post
emy’s vulnerability while crossing by using artillery and Platoon Ambush) were adapted with permission
to prevent it from concentrating forces on either from among the tactical decision games presented
side. To guard against the possibility that predictive in the Marine Corps Gazette. A participant's task
and proactive tactics fail to achieve their purpose, was to read the scenario, to develop operations or-
the plan is further elaborated to include a flexible, ders, explain the orders, describe alternative plans,
reactive orientation in case the enemy does some- list the strengths and weaknesses of all plans, give
thing unexpected. The result is a mutually suppor- reasons for choosing the preferred plan, and indicate
tive network in which proactive tactics are utilized how weaknesses in the preferred plan are addressed.
Dependent measures included references to high considerations might justify the decision to attack
level purpose, references to proactive, predictive, Sanna’s Post in terms of the larger battalion fight,
reactive or predictive-reactive (contingency) plans, e.g., to prevent the enemy using the tanks posi-
and the actual tactics adopted. Unfortunately, results tioned there against the battalion, to reduce the en-
for one of the scenarios (Platoon Ambush) showed emy’s logistical support for the fight against the bat-
very little variability among students (and no sig- talion, or to force the enemy to orient away from the
nificant effects), possibly due to ceiling effects. battalion. Relatively short-term proactive tactics
might also be adopted in an attack, e.g., use of sur-
Results prise, speed, and multiple directions of attack by the
company to create confusion at Sanna’s Post. Train-
Reaction. In the Sanna’s Post scenario, the ing significantly increased the proportion of officers
company mission is to guard the battalion’s left who used proactive plans in their favored course of
flank, and it is necessary to deal with the unex- action (F(1,50) = 5.016; p = 0.030). A breakdown in
pected presence of enemy vehicles at a small set- terms of short-term versus long-term proactive ele-
tlement, Sanna’s Post, that sits on a tactically im- ments reveals that the largest effect of training was
portant road. The first question, therefore, con- on longer-term proactive planning (F(1,50)= 9.584;
cerned the reactive time orientation: Did officers p = 0.003) – but the difference was only marginally
adopt any action at all with respect to the unex- significant.
pected enemy at Sanna’s Post (down to and includ- Predictive-reactive. The predictive-reactive
ing a simple decision to keep an eye on them)? time orientation (i.e., anticipating specific possibili-
There was a significant increase in attention to ties and making contingency plans) might also be
Sanna’s Post after training (F(1,50)=18.041; p = adopted at either a short-term or a long-term level
0.000). Before training only 10.5% of the officers of planning. Long range plans might include con-
took any action, no matter how minor, with regard tingencies for taking over the battalion main effort
to Sanna’s Post, while 63.6% of the officers took in case it became necessary. Short-range plans
some action after the training. might include contingencies for acting in case of
Purpose. A good argument could be made that specific enemy movements on the road through
the company should take the initiative via an offen- Sanna’ Post, or for deploying different types of
sive action against the enemy at Sanna’s Post, to forces in a company attack on Sanna’s Post. Train-
protect the battalion’s left flank. On the other hand, ing significantly increased the proportion of officers
the impact of such an initiative on coordination with who incorporated branches or contingencies into
the battalion, hence, on success over the longer- their plans, from 27.3% to 57.9% (F(1,50)= 5.058;
term, needs to be critically considered. Guarding the p =0.029). The size of the effect was larger for
flank of the battalion implies a longer term purpose short-term than for long-term contingencies, but
of staying ready to move when the battalion moves. there was no significant difference.
Becoming bogged down in a fight at Sanna’s Post Tactics. We also looked at the effects of train-
may make this difficult and cause the battalion’s ing on the specific tactics that participants adopted.
flank to be exposed in the future. After training, ref- Officer’s plans were categorized as either including
erences by officers to this longer-range or higher- or not including three non-mutually exclusive ele-
level purpose increased significantly in their ac- ments. All three showed significant changes after
counts of their reasoning (F(1,50)= 7.024; p = training. There was an increase from 15.2% to
0.011). In the pretest only 3% of the participants 47.4% in the use of artillery against Sanna’s Post
mentioned the importance of maintaining contact (F(1,50 ) = 6.969; p = 0.011); an increase from
with the battalion as it moved, whereas 26.3% did 24.2% to 57.9% in the (possibly contingent) use of
so after training. a ground attack against Sanna’s Post (F(1,50) =
Proaction. A proactive time orientation can be 6.386; p = 0.015); and an increase from 39.4% to
adopted at any level of planning. For example, in 78.9%, in the defense of two river fords from high
the Sanna’s Post scenario, longer-term proactive ground against enemy reinforcement of the battalion
(F(1,50) = 8.546; p = 0.005). A correlational analy- the initiative in shaping enemy decisions, and (ii) to
sis (using Pathfinder) showed that the effects of ensure that such shaping served the higher-level
training on all three tactical elements, as well as on purposes of the organization.
the use of contingency plans, was mediated by the
direct effect of training on proactive planning.
Prediction. Training had no effect on the use of ACKNOWLEDGMENTS
the predictive time orientation, e.g., calculating the
likelihood of success against the forces in Sanna’s This research was sponsored by Contract No.
Post, or predicting the intent of the enemy there. DASW01-97-C-0038 with the Army Research Insti-
tute, Fort Leavenworth, KA. We are grateful to Dr.
SUMMARY AND CONCLUSION Sharon Riedel, who was the scientific monitor of
the project, and to LTC Billy Hadfield for his in-
We did not assume that there was a single right valuable help. Others who contributed to this work
answer in the Sanna’s Post scenario. Rather, our are Terry Bresnick, Bryan B. Thompson, and Jonah
interest was on the thought processes that lay be- I. Cohen.
hind participants’ solutions. Actions based on pre-
diction may get a jump on the enemy, but can fail if REFERENCES
predictions are wrong. One way to deal with this
risk is to take even more initiative, by attempting to Brown, Lesley (Ed.) (1993). The new shorter Ox-
proactively influence, rather than simply predict, the ford English dictionary. Oxford: Clarendon Press.
enemy’s future actions (e.g., drive the enemy out of Cohen, M.S., Freeman, J.T., & Wolf, S. (1996).
Sanna’s Post before they can intervene in the battal- Meta-recognition in time stressed decision mak-
ion fight). But initiative creates another risk: dimin- ing: Recognizing, critiquing, and correcting. Hu-
ished coordination with other friendly forces (e.g., man Factors, 38(2), pp. 206-219.
loss of contact with the battalion). And this risk, in Cohen, M.S., Thompson, B.B., Adelman, L., Bres-
turn, can also be mitigated. One way is to heavily nick, T.A., Shastri, L. & Riedel, S. (2000). Train-
weight the purposes of higher level units (e.g., what ing critical thinking for the battlefield. Arlington,
is the long-range benefit or cost to the battalion of VA: Cognitive Technologies, Inc.
this action). The other way is to use contingency Cohen, M.S., & Thompson, B.B. (2001). Training
plans, to prepare ahead for situations in which plans teams to take initiative: Critical thinking in novel
may fail. situations. In E. Salas (Ed.), Advances in Cogni-
Training was designed to help officers balance tive Engineering and Human Performance Re-
the benefits and the risks of taking initiative in un- search, Vol. 1. JAI.
certain situations. The training was based on critical March, J.G. (1996). Exploration and exploitation in
incident interviews in which we found that more organizational learning. In M.D. Cohen & L.S.
experienced Army officers were more likely to plan Sproull (Eds.), Organizational learning (pp. 101-
proactively and to focus on higher-level purposes 123). Thousand Oaks, CA: Sage Publications.
than less experienced officers. Training succeeded
in making students more like experienced officers
in these respects for a scenario in which the initial
level of performance was not too high. It increased
the frequency with which participants used a proac-
tive time orientation to influence the enemy,
thought about the impact of the initiative on higher-
level purposes, and made contingency plans in case
things went wrong. Training also increased the use
of three key tactical elements. In sum, the primary
effects of training were (i) to get participants to take