Pattern Recognition in Child Welfare
(Originally published November 2010)
In Streetlight and Shadows: Searching for the Keys to Adaptive Decision Making, Gary Klein expands on the naturalistic description of expert decision making he began in Sources of Power: How People Make Decisions (1999). Klein’s descriptions of expert decision-makers under time pressures, or when faced with complex or ambiguous situations for which rules have not been developed, emphasize experts’ reliance on intuitive pattern recognition rather than logic, statistics and/ or a rational weighing of pros and cons regarding possible courses of action. Sources of Power utilizes studies of firefighters, submarine commanders and chess masters who, under severe time pressures, almost instantly recognize typical patterns along with anomalous and unexpected variations from these patterns, engage in mental simulation of possible courses of action, identify ‘good enough’ solutions to problems or challenges and often depend on tacit knowledge that they can not explain to others.
Klein’s argument is that logic models of rationality do not take account of how experts behave when there is no time for careful reasoning, or when reason is unable to cope with unanticipated complexity. Fans of the Bourne movies (The Bourne Identity and others) will understand the fascination of a character with amnesia regarding his identity and his past, but who can instantly and effectively deal with threats and the complexities of the spy game. Bourne has tacit knowledge that he has no memory of acquiring; he notices patterns and their meaning, and he has a large repertoire of effective responses at the ready. Furthermore, Bourne’s responses are lighting fast and automatic, a great fantasy for those of us who have to struggle to complete simple tasks.
Streetlights and Shadows challenges ten claims about how to think effectively in lengthy arguments filled with examples from a variety of professions, especially piloting, weather forecasting and military command. These claims are:
Teaching people procedures helps them perform tasks more skillfully.
Decision biases distort our thinking.
Successful decision makers rely on logic and statistics rather than intuition.
To make a decision, generate several options and compare them to pick the best one.
It’s bad to jump to conclusions – wait to see all the evidence.
To get people to learn, give them feedback on the consequences of their actions.
To make sense of a situation, we make inferences from the data.
The starting point for any project is to get a clear description of the goal.
Our plans will succeed more often if we identify the biggest risks and find ways to eliminate them.
Leaders can create common ground by assigning roles and setting ground rules in advance. (p.8)
Klein finds numerous exceptions to each of these rules and reviews evidence that often severely limits the scope of their application. Again and again, he utilizes naturalistic descriptions of skilled decision makers and of expertise in various professions to counter overreaching claims for analytical approaches to decision making he believes are uncritically accepted by most professionals and other educated persons. For example, Klein acknowledges the usefulness of procedural checklists, but finds that pilots and other professionals often admit using unwritten meta - rules or intuition for deviating from standard procedures. Checklists are most effective when used in combination with judgment and the intuition of skilled professionals, not in place of them, according to Klein.
Klein believes that dependence on procedures and decision tools can undermine the development of expertise. He states that “A number of studies have shown that procedures help people handle typical tasks, but people do best in novel situations when they understand the system they need to control. People taught to understand the system develop richer mental models than people taught to follow procedures.” (p.23) It is the intersection of pattern recognition with the understanding provided by “richer mental models” that gives Klein’s account of expertise a powerful application to child welfare.
I began to be disenchanted with risk assessment models (both actuarial models and those developed in other ways) over a decade ago when it became apparent that their use in child welfare settings had become a substitute for understanding the dynamics of child abuse and neglect. The emphasis of risk assessment models on prediction in lieu of explanation, in combination with procedural approaches to decision making, has led to a “dumbing down” in child protection when caseworkers responsible for investigations and assessments of families approach these demanding tasks with weak or non - existent conceptual frameworks for understanding what they encounter in troubled families.
As it turns out, the risk assessment tools widely used in child welfare have modest predictive powers when applied to specific families, primarily because of the large percentage of “false positives”, i.e., families classified as high risk due to clusters of risk factors or extensive histories of child maltreatment, but who are not found to abuse or neglect their children. Risk assessment models are not (as once claimed) adequate alternatives to clinical judgment; and the compiling and contrast of risk factors and protective factors is not an adequate basis for case planning. Actuarial risk assessment models do a better job of classifying cases according to risk categories, i.e., high, moderate and low, than models developed in other ways; but the decisions that can be reasonably made on the basis of a case’s risk category are severely limited, e.g., to open or close the case following investigation.
The expertise that is needed in child protection is much along the lines that Klein has described: “Every type of expert we have studied has built up a repertoire of patterns to quickly make sense of what is happening. These patterns aren’t facts, rules or procedures. They are based on all the experiences and events the experts have lived through and heard about. They are the basis of intuitions.” (p.43) What Klein does not say because of his fascination with intuition is that pattern recognition can be taught, and that experts can convert tacit knowledge into analytical understanding. A professional who is outstanding at her / his job will not be a good teacher of novices unless they can break down what they do into specific skills and attitudes. A high level of tacit knowledge is impressive to watch, but it is not the basis for a training curriculum.
One potentially effective approach to teaching pattern recognition in child protection and child welfare is to find or develop operationally useful typologies of child maltreatment. In 1990, Carol Bowdry, then a training manager in Dallas, Texas, published an operationally useful typology of physical abuse in Child Welfare:
Battered child syndrome
Escalating cycle of injury
Excessive punishment with rejection
Misguided attempts at education of children
A few years ago I added "Intimate terrorism" to this list to account for the subset of domestic violence cases in which one parent, almost always the male partner, terrorizes his female partner and their children through calculated extreme violence, threats and psychological aggression, including large doses of humiliation.
Competent decision makers in child welfare agencies and courts quickly grasp the difference between a typical case of excessive spanking and torture or between torture and misguided attempts at education. Bowdry’s article spells out the operational meanings of these classifications for safety decisions and, in some instances, for treatment planning. Understanding that torture is ‘high risk’ is not a substitute for grasping the malignant implications of physical abuse that has become a thoughtful deliberate effort to inflict pain, usually justified as an essential phase in a power struggle with a child parents feel they must win at all costs.
I have developed two operationally useful typologies for neglect, one based on referral histories, the other that identifies chronic conditions that undermine parenting abilities:
Referral Histories Chronic Conditions
Situational neglect Substance abuse with depression
Sporadic neglect Substance abuse with anti–social features
Chronic neglect Substance abuse as a means of self destruction
Chronic maltreatment Mental illness with periodic breakdowns
Severe cognitive impairment with deficiencies in nurturance
Intergenerational poverty combined with social isolation
Caseworkers who understand the differences between situational neglect and chronic neglect will engage in different types of assessment processes and offer different types of services for families depending on their referral histories and the presence or absence of chronic conditions that diminish parenting abilities. Straightforward discussions regarding allegations of child abuse or neglect and brief, often poverty related services, are reasonable responses to situational neglect, but indicate a caseworker’s denial or lack of understanding in sporadic neglect, chronic neglect or chronic maltreatment. Parents with anti – social features require a special emphasis on structure and consequences, and are often surprisingly responsive to a calm, just and dispassionate application of rules. They may also “bite off their nose to spite their face” when engaged in a conflict with authority figures. Caseworkers who ignore parents’ criminal histories, or become involved in bitter highly personal power struggles with persons who carry a grudge toward authority figures do so at their own risk.
The development and testing of operationally useful typologies is one approach to teaching pattern recognition, but not the only one. Caseworkers need to be able to recognize the indicators of phenomena that they will encounter frequently in child welfare work: relapse (erratic compliance with mandated services, withdrawal from professionals, excuses and lies in defense of bad behavior), parental ambivalence regarding reunification (self sabotage, highly sensitive or avoidant responses to probing questions, flip flops regarding life path and goals) , children’s emotional conflict following visitation with parents ( cranky behavior, sleep problems), decision making bias (refusal to consider new information, distortion or denial of evidence, seeking confirmation of one’s views) and so forth. The most effective approach to teaching pattern recognition in training programs is to present caseworkers with case scenarios that include clusters of indicators which, taken together, present coherent patterns. It can be boring to be lectured about relapse, counter - phobic reactions, caseworkers’ enmeshment with parents, effects of traumatic stress or early severe deprivation of nurturance and basic care on child development, but it’s exciting to recognize these phenomena not as abstractions but as clusters of behaviors and attitudes that present themselves in many various ways.
There is a widespread perspective that using typologies and recognizing typical patterns of behavior depersonalizes individuals and families. It’s possible to use DSM diagnoses in ways that practically identify suffering persons with diagnostic entities. It’s possible to use categories of moral judgment in ways that prevent understanding and empathy. Typologies and other approaches to classification can be used in ways that are injurious to persons social workers are attempting to help. Theories can blind as well as uncover deep causal structures. However, any approach to interacting with and understanding families in child protection carries risks. Engagement skills can lead to enmeshment with parents rather than productive partnerships between caseworkers and family members. Strengths based practice used foolishly can lead to fatuous assessments that minimize immediate safety threats and risks of cumulative harm to children’s emotional and social development. The immediate challenge facing child welfare agencies is to strength caseworkers’ family engagement skills, while also improving the conceptual frameworks that guide safety planning and therapeutic interventions.
Gary Klein’s work challenges the idea that there is a substitute for expertise. He writes:
“Several claims (listed above) suggest ways that we can replace expertise – with procedures, with explicit knowledge that can be added into mental storehouses, with methods to calculate the best decisions, with emphasizing logic and analysis over biased intuitions, with systematic steps to convert data into knowledge, with computational decision aids, with sufficient data, with managing by objectives, with strategies for controlling risks." However, he continues, “the different ways of replacing expertise will not work well in the shadowy world of ambiguity and complexity” (p. 297) in which goals conflict and procedural compliance will not achieve results.
Klein’s description of experts is that they are outstanding “detectors” who notice cues invisible to most of us. “They can make discriminations that most of us can’t make and … they pick up cues, patterns and trends that go undetected.” (p.297) Experts can combine analytical methods and intuition, research findings and practice wisdom, and adapt to circumstances that policy and procedural manuals have not anticipated, according to Klein.
The child welfare field has spent much of the past 20 years attempting to standardize expertise in assessment tools and procedural guidelines without developing experts on Klein’s model. Nevertheless, most child welfare agencies have experts – especially at the supervisory level – who can serve as role models for training programs. It is time to complement the effort to standardize decision making with initiatives that develop critical thinking skills, teach pattern recognition, and that contain a renewed appreciation of practitioners’ need to understand child maltreatment as well as predict it.
Bowdry, Carol, “Toward a Treatment Relevant Typology of Physical Abuse,” Child Welfare , 1990.
Klein, Gary, Sources of Power: How People Make Decisions, The MIT Press, 1999.
Klein, Gary, Streetlights and Shadows: Searching for the Keys to Adaptive Decision Making, The MIT Press, 2009.
Loman, Anthony, Frequently Encountered Families , Institute for Applied Research, 200 .
Shlonsky, Aron and Wagner, Denis, Wilson, Dee and Horner, William, “Chronic Neglect: Needed Developments in Theory and Practice,” Families in Society, November / December, 2005.