finaldraftforPCL9


 1 

 
 
 
 
 
 
 
 
 
 
 
 
 
 

Achieving Cumulative Progress in Understanding Crime: 
Some Insights from the Philosophy of Science 

 
 
 

“Understanding Crime” special issue, Psychology, Crime & Law 
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 



 2 

 
Abstract 

Crime is a serious social problem, but its causes are not exclusively social. There is growing 

consensus that explaining and preventing it requires interdisciplinary research efforts. Indeed, the 

landscape of contemporary criminology includes a variety of theoretical models that incorporate 

psychological, biological and sociological factors. These multi-disciplinary approaches, 

however, have yet to radically advance scientific understandings of crime and shed light on how 

to manage it. In this paper, using conceptual tools on offer in the philosophy of science in 

combination with theoretical work represented in this special volume of Psychology, Crime and 

Law, I provide some perspective on why explanatory progress in criminology has remained 

elusive and evaluate some positive proposals for attaining it.       

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 



 3 

Introduction  

Criminal behaviors—murder, rape, incest, arson—are structurally complex phenomena 

insofar as investigators may probe various parts of the fabric of the world to locate their causes. 

Since the scientific search for the causes of crime began in the 18th century (Rafter, 2011), 

criminology has expanded to include psychological and biological approaches in addition to 

sociological and legal ones. Today, there is growing consensus among contemporary 

criminologists that progress in understanding crime will not be possible within the confines of 

any single scientific discipline. Rather, explaining and preventing criminal behavior requires an 

integrative approach that unifies insights from biology, genetics, sociology, psychology, 

neuroscience and related fields. Indeed, when we survey the landscape of contemporary 

criminology we find a plurality of interdisciplinary theoretical and explanatory approaches to 

crime. To date, however, this pluralism has not radically advanced scientific understandings of 

criminal behavior nor shed adequate light on how to prevent it.  

In this paper, I offer one perspective on why interdisciplinary progress in criminology has 

remained elusive and evaluate some proposals for facilitating it. Part of the problem, I suggest, 

partially on the basis of the disciplinary review papers contained in this special issue, is that what 

we find within and across different scientific fields that study the causes of crime is unbridled 

rather than coordinated pluralism: investigators working within the same and across different 

fields of criminology have different ways of identifying, conceptualizing and classifying criminal 

behaviors and their causes as well as different methods and evidential standards for investigating 

them. These differences, rather than facilitating cumulative progress, serve as barriers to 

effective interdisciplinary communication that is necessary to facilitate causal discovery. The 



 4 

relevant question thus becomes, how might practices within and across these different areas of 

science change to make an interdisciplinary understanding of crime an attainable goal?  

In order to answer this question in this paper, I begin, in section 1, by introducing some 

conceptual tools from the philosophy of science. These tools merely serve as heuristics for 

understanding the structure of those areas of science that are interested in understanding crime 

and how progress given this structure may be achieved. In section 2, I consider specific cases in 

which scholars have tried to put forward integrative accounts of crime. I apply the conceptual 

tools to evaluate what is going on in these cases and relate this analysis to how practitioners in 

these fields themselves conceive of the problems. I go on to note that the kind of obstacles to 

progress that what we find in criminology are not unique to it; investigators undertaking research 

on mental illness face similar challenges. I end by evaluating several proposals offered in this 

special issue for securing interdisciplinary progress in crime research and supplement them with 

some additional guidelines I regard as important for putting the various fields of criminology on 

a trajectory for success. 

 

Conceptual tools 

One aim of this special issue of Psychology, Crime and Law on “Understanding Crime” 

is to address the question of how those disciplines engaged in investigating the causes of 

criminal behavior may make progress in understanding and explaining it. In recent years, a small 

number of scholars involved in the empirical study of crime have looked to philosophical models 

of scientific progress for normative guidance (Durrant and Ward, 2015; Ward, Wilshire, and 

Jackson, 2018). In this section, I describe a set of conceptual tools on offer in philosophy of 

science for thinking about scientific progress, especially in areas of science that investigate 



 5 

complex phenomena. I then use these tools, in the next section, to evaluate some representative 

scientific approaches to crime. This analysis is intended to be a complement to that theoretical 

work in the sciences of criminology showcased in this volume (Day, Tamatea, and Gaia, 2019; 

Durrant, 2019; Fortune and Heffernan, 2019; Dixon, Harkness, and Wegerhoff, 2019; Raine and 

Ling, 2019; McGee and Farrington, 2019; Ward and Carter, 2019; Weaver, 2019).  

Much contemporary work in philosophy of science may be viewed as a reaction to ideas 

about science put forward by the logical positivists in the first half of the 20th century 

(Oppenheim and Putnam, 1958; Nagel 1961). One fundamental positivist idea is that phenomena 

and the scientific study of phenomena may be subdivided into levels with each branch of science 

(e.g., sociology, psychology, physics) situated at a single “level of analysis” taking one specific 

“level of organization” of matter (e.g., social groups, individuals, atoms) as its focus of inquiry. 

Levels of analysis are organized hierarchically, lower levels study the parts (e.g., individuals) of 

the wholes (e.g., social groups) under study at a higher level. The hierarchical ordering relation 

among levels is from complex to simple, from large to small, from less to more fundamental, or 

from whole to part. Each level may be treated as discrete and theoretically independent from 

another and investigators at each level deploy particular terms to individuate those kinds of 

entities and properties that are "fundamental" to their science and its associated theory. On this 

picture, sociology studies social groups and is at the topmost level, psychology studies 

individuals that comprise social groups and is one level down, biology is at the next level down 

insofar as it studies the somatic parts of individuals, and physics, because it studies the most 

fundamental units of matter of which everything else is composed, is at the lowest level.  

The logical positivists believed that scientists should seek unified understandings of 

phenomena (Nagel, 1961; Oppenheim and Putnam, 1958). On one model proposed by Ernest 



 6 

Nagel, unification was to be achieved by the process of intertheoretic reduction. The basic idea 

is that progress in science occurs when a theory that corresponds to one branch of science 

situated at one level of analysis is reduced to a theory corresponding to a branch of science 

situated at a lower-level of analysis. Intertheoretic reduction requires the satisfaction of two 

conditions: connectability and derivability. According to the connectability condition, the terms 

in the reduced theory must be connected to the terms in the reducing theory by means of bi-

conditional bridge laws that specify an identity relation between the terms and their referents. 

The derivability condition, in contrast, requires that the reducing theory be capable of explaining 

all the phenomena the reduced theory was originally able to explain. The classic example is the 

reduction of thermodynamics to statistical mechanics; not only can the latter theory purportedly 

explain the same phenomena that former could, but terms contained in thermodynamic theory 

(e.g., “temperature of a gas”) could be shown to be bidirectionally equivalent to terms in 

statistical mechanics (e.g., “mean kinetic energy of molecules comprising the gas”). The ultimate 

aim of all of science, on Nagel’s model, was the gradual reduction of all higher-level sciences to 

fundamental physics. 

Philosophers of science still traffic in talk about levels of organization and analysis but 

recognize both as imperfect heuristics for understanding the structure of the world and science 

(Bechtel, 1986; Machamer and Sullivan, 2001; Potochnik, 2010; Potochnik and McGill, 2012). 

Few, if any, however, endorse the idea that progress in science involves intertheoretic reduction. 

In fact, most work in philosophy of science in the latter half of the 20th century and early 21st 

century has been directed at critiquing and proposing alternatives to Nagel’s model for 

understanding the structure of science and the nature of scientific progress.  



 7 

In the 1970’s, for example, philosophers of science suggested that the appropriate unit of 

analysis for understanding the organization of different areas of science was not “level” but 

rather, some alternative such as field (Darden and Maull, 1977), domain (Shapere, 1974), or 

discipline (Bechtel, 1986). Moreover, the kinds of interactions regarded as occurring within and 

among different scientific fields involved something other than establishing reductive 

relationships between different theories. Fields of science came to be regarded as consisting of 

individual investigators who shared explanatory problems in common. Sometimes, when 

confronted with an intractable problem, scientists working in one field turned to scientists 

working in another for guidance, borrowing an experimental method or a theoretical tool 

(Bechtel, 1986; Darden and Maull, 1977). There were also instances in which scientists working 

in different fields joined forces to solve explanatory problems of mutual interest and theories in 

those fields appeared to “co-evolve” (Churchland and Sejnowski, 1988). By considering case 

studies particularly from the biological and mind-brain sciences, philosophers explained that 

interdisciplinary relationships between different areas of science could be many and varied 

(Bechtel, 1986; Bechtel and Richardson, 1993; See also Mitchell, 2003, Chapter 6 on 

“integrative pluralism”). Scientific progress involved far more complex interactions within and 

across fields than the logical positivists had appreciated.  

In conjunction with this new appreciation of the structure of science, philosophers became 

concerned with goals in science other than theory development. In the late 1950s, the nature of 

scientific explanation became the new focus as philosophers began to critically evaluate the 

deductive-nomological (D-N) model of scientific explanation (Hempel and Oppenheim, 1948). 

On this model, explanations were conceived as arguments; the premises consisted of statements 

of laws and initial conditions that were taken to logically imply the observations to be explained 



 8 

or predicted. Critics of the model claimed not only that it got explanation in physics wrong, but 

also that it was descriptively inadequate when it came to sciences like biology and psychology 

that sought to explain phenomena of interest in the absence of laws (Bechtel, 2008; Bechtel and 

Richardson, 1993; Cummins, 1975, 1983; Fodor, 1968; Salmon, 1989). To this end, in the latter 

half of the 20th century, alternative accounts of scientific explanation were proposed. One 

alternative, mechanistic explanation bears directly on questions of explanatory progress and the 

feasibility of developing integrative explanations for complex phenomena like crime (Bechtel, 

2007; Bechtel and Richardson, 1993; Craver 2007; Machamer, Darden, and Craver, 2000; 

Piccinini and Craver, 2011; See also Stinson, 2016; Stinson and Sullivan, 2017).  

Mechanisms are causal systems that contain entities/components and their 

activities/processes that are organized so as to give rise to the behavior of that system (Craver, 

2007). A mechanistic explanation describes how those parts, namely, entities/components and 

activities/processes and their spatiotemporal organization, actively give rise to that behavior. To 

take an example, a mechanistic explanation of the depolarization of the neuron would include the 

coordinated opening and closing of sodium and potassium ion channels in the cell membrane, 

which permit changes in the flow of ions across the membrane.    

Descriptions of mechanisms are typically put forward to explain complex phenomena (e.g., 

criminal behavior, mental illness, cognition) that are a consequence of entities and activities 

operating at multiple different levels or scales of organization within a nested hierarchy. 

Mechanistic explanation is thus consistent with an account of scientific progress in which 

different fields of science that investigate different levels of organization each contribute to 

filling out the multi-level details of the explanation. Findings from individual fields, then, “are 

used, like the tiles in a mosaic, to elaborate the mechanism sketch, showing piecemeal which 



 9 

components and properties are relevant to the mechanism and how those components are 

organized spatially, temporally, and actively within the mechanism” (Craver 2007, 524). To take 

Craver’s example, the mechanism of spatial memory in rodents is widely described as involving 

the formation of a spatial map in the hippocampus, which is brought about by changes in 

synaptic plasticity in area CA1 of the hippocampus, which is mediated by activation of N-

methyl-D-aspartate receptors in area CA1 neurons. Findings from neurophysiology, cellular and 

molecular and behavioral neuroscience have all contributed to the development of this 

mechanistic explanation (See also Sullivan, 2010, 2016).     

Gualtiero Piccinini and Carl Craver (2011) provide a more explicit statement of how 

mechanistic explanatory integration works by considering the relationship between explanations 

in psychology and neuroscience. Although psychology and neuroscience are regarded as distinct 

scientific fields, Piccinini and Craver argue that they are not explanatorily autonomous; while 

both areas of science aim to explain cognitive capacities, only neuroscience is successful insofar 

as it identifies both the functional and the structural details—the activities and the entities—of 

the physical systems that realize psychological processes and properties. Piccinini and Craver 

may be described as conceiving of the two forms of explanation as situated at different points on 

an explanatory completeness continuum. Functional analyses or “mechanism sketches” lie at one 

end; complete mechanistic explanations lie at the other. Once neuroscience fills in “the structural 

aspects that are missing from a functional analysis,” it “turns into a more complete mechanistic 

explanation” (Piccinini and Craver, 2011, 308).  

Mechanistic explanatory integration has obvious descriptive advantages over intertheoretic 

reduction as a model for understanding scientific progress. First, it is applicable to areas of 

science that lack bona fide theories. Second, explanatory progress is achieved as different areas 



 10 

of science contribute findings to common explanations of common phenomena while 

maintaining their status as autonomous disciplines or fields. The model also can accommodate a 

variety of types of relationships that may exist between different scientific disciplines or fields, 

although its advocates particularly value interactions that facilitate explanatory unification.   

Despite certain advantages over the Nagelian model, however, mechanistic explanatory 

integration has at least one important limitation.  In order for two theories from different areas of 

science to be reductively related to each other, Nagel recognized that terms designating the kinds 

to which one theory referred had to have (roughly) the same referents as the terms in the other. 

The two terms had to be “connectable” via bridge laws that specified their bidirectional 

equivalence. The model of mechanistic explanatory integration, in contrast, leaves mysterious 

precisely what is being fitted together when “results” from the same and different fields are fitted 

together in mechanistic explanations. Mechanistic explanations involve terms that designate 

entities, activities and phenomena to be explained. A prerequisite for results arising from the 

same and different areas of science to fit into a single mechanistic explanation of a common 

phenomenon is for the terms designating causal variables and explanatory targets to have the 

same referents. The meanings of the terms must be stable within and across different fields of 

science that contribute to the mechanistic explanation in order for integration to be possible 

(Sullivan, 2016).  

If I am correct, mechanistic explanatory integration needs to be supplemented with some 

tools for thinking about other aspects of scientific practice that are important to developing 

successful explanations (Sullivan, 2009, 2015). For reasons of space, I cannot discuss the 

different typesof practices that may be relevant to the development of integrative multi-level 

explanations of crime, so I focus here on conceptual practices. Methodological practices are 



 11 

certainly also relevant, as differences in evidential standards upheld within and across different 

fields of science may also impact the feasibility or legitimacy of explanatory integration. The 

recent replication crisis in science has called attention to the fact that sometimes investigators 

working within the same field do not maintain the same standards. (See Baron and Sullivan, 

2018; Haig, 2019; McGee and Farrington, 2019; Sullivan, 2015).  

By conceptual practices, I mean to capture the ways in which investigators deploy concepts 

so as to pick out (a) phenomena that are the targets of explanation as well as (b) the causal 

variables they identify to explain those phenomena. Prior to a scientist running an experiment, 

she will have grouped together instances of what she takes to be the same phenomenon or 

variable under a concept or construct. She may rely on how other investigators in their field 

define the concept, but she may also define it slightly differently. Folk psychological 

understandings of the concept may shape the definition that she provides (Sullivan, 2014) or she 

may instead be influenced by how practitioners working in a different or related scientific field 

understand the concept. Some examples of constructs in criminology include: antisocial 

behavior, aggression, risk factor, crimogenic need, interpersonal violence, fear-conditioning, 

well-being and desistance. These constructs originate with a concept that an investigator 

associates with certain observations, which serves as a basis for theory building and the 

development of procedures for detecting instances of them in experimental settings.  

Once an investigator has selected a capacity or property of interest, which is designated by a 

construct, she specifies a set of procedures for producing, detecting, and/or measuring an 

instance of it in an experiment or research study. In other words, she operationally defines the 

term (Chang, 2010). For example, an investigator might operationally define a construct like 

antisocial potential as an individual’s propensity to commit antisocial acts or “prosocial 



 12 

behavior” in terms of degree of attachment to one’s parents (McGee and Farrington, 2019). 

Ideally, she aims for the set of procedures that she selects to detect instances of the construct to 

have a high degree of construct validity. In other words, she aims for the match between what 

she investigates in her experiments and the instances of the phenomena or property she takes to 

be grouped under the construct to be valid (Cronbach and Meehl 1955). Experiments may have 

anywhere from a low to high degree of construct validity. The higher the degree of construct 

validity, the closer the match between the variables or effects under study in the research study 

and the variables or phenomena designated by the constructs.  

Several features of conceptual practices in science are relevant to the issue of explanatory 

integration as well as to the development of theories that incorporate concepts from different 

areas of science. First, investigators working within the same field may not necessarily agree 

about how to define a given construct. Second, they may disagree about the best procedures for 

producing, detecting and/or measuring instances of those constructs in experimental settings. 

Third, investigators may, in light of their research findings, revise their understandings of 

constructs or the procedures they use for producing, detecting and measuring instances of them 

across experiments, irrespective of how other members of their field understand those terms. 

Fourth, these features of conceptual practices that may apply within a single scientific field may 

also hold true across fields. In other words, there may potentially be variation with respect to (1) 

how investigators from different fields use terms and (2) what procedures they use to produce, 

detect and measure instances of the kinds the terms pick out. 

 Armed with these conceptual tools, I want to consider several examples of integrative 

theories or mechanistic explanations in contemporary criminology with an eye towards 

identifying obstacles to interdisciplinary progress in understanding criminal behavior. For 



 13 

reasons of space, this analysis is not exhaustive; I do not consider all of the different fields or 

details of those fields that comprise contemporary criminology. What I offer is at best a first pass 

at pinpointing the precise places where obstacles to explanatory progress emerge within and 

across fields. It is to a large extent based on the meta-analytic work undertaken by scholars in 

this special issue who have sought to engage in a critical examination of their own fields’ 

contributions to understanding and explaining crime. The philosophical analysis I provide is 

intended as a complement to this field-specific analytic work.  

 

The Conceptual, Theoretical and Explanatory Landscape in Criminology 

In order to appreciate why criminologists believe that explaining criminal behavior 

requires an interdisciplinary approach, it is worthwhile to consider several examples. Burns and 

Swerdlow (2003) describe the case of a forty-year-old male schoolteacher who began making 

sexual advances towards his stepdaughter and soliciting prostitutes. He had no prior history of 

engaging in these behaviors. An explanation for these changes in his behavior may have 

potentially appealed to (a) the act of his spouse rejecting his sexual advances, (b) his feelings of 

inadequacy in response to this act, (c) a loss of self-control, or (d) a brain abnormality. Clearly, 

these different variables may have been related to each other in complex ways and interacted 

dynamically in the production of the behavior. Ideally, any explanation provided for his behavior 

would track the real as opposed to the hypothetical causes. It was determined that the man did 

have a tumor that was large enough to disrupt normal functioning of the frontal lobe and 

hypothalamus, and that its growth correlated with the onset of his criminal behavior.  

In another case, Andrew Goldstein, who had a history of schizophrenia, threw Kendra 

Webdale to her death in front of a New York City subway train (People v Goldstein, 2004). An 



 14 

explanation for this violent act may have posited (a) a history of mental illness, (b) an 

abnormality in his brain, (c) his feelings of social rejection on the basis of his mental illness, (d) 

failure to take his medication, (e) his inability to differentiate right from wrong, or (f) his lack of 

self-control. Again, an explanation for Goldstein’s behavior could have appealed to any and all 

of these variables and their dynamic interactions. Although, Goldstein originally claimed he was 

unable to recognize the difference between right and wrong, he later admitted that he willingly 

and knowingly committed the crime. 

The psychology of criminal conduct has as its aim to explain the onset and maintenance 

of anti-social behavior, which includes actions like those described above that negatively impact 

the well-being of others. The field at one time relied exclusively on sociological theories of 

crime (Andrews, Bonta, and Wormith, 2011; Fortune and Heffernan, 2019), but as psychological 

approaches to the field developed, investigators recognized causal variables internal to 

individuals who committed crimes (e.g., psychological and biological risk factors) were as 

important as external factors (e.g., social, cultural and familial environments) in understanding 

their behavior. During the past several decades, as the field has grown, an interfield theory 

known as the General Personality and Cognitive Social Learning Theory of Criminal Conduct 

(GPCSL) has been developed and refined. The aim of the theory is to explain antisocial behavior 

by means of a broad “range of psycho-social-biological factors which contribute to the likelihood 

of antisocial behavior occurring and its maintenance over time” (Fortune and Heffernan, 2019, 

XX). Using this theory and its related components (the Central Eight risk factors, the Risk Need 

and Responsivity Model and related principles (Fortune and Heffernan, 2019)) investigators may 

appeal to a variety of different psychological, social, environmental and biological risk factors to 



 15 

explain why an individual committed a crime and to predict the likelihood that they will engage 

in offending behavior in the future.  

A variety of different constructs are associated with the GPCSL framework. “Risk factor” 

is itself a construct, and different investigators define the concept in different ways (Fortune and 

Heffernan, 2019). Constructs designating the Central Eight risk factors for anti-social behavior 

include: (1) a history of antisocial behavior, (2) antisocial personality traits, (3) criminal 

thinking, (4) criminal associates, (5) substance abuse,(6) family/marital problems,(7) poor 

work/school performance and (8) lack of prosocial leisure activities. Other terms have been 

introduced into the literature that subdivide the Central Eight into subconstructs. The list of 

subconstructs includes: intimacy deficits, self-regulation problems, emotional congruence with 

children, negative affect and poor coping skills, to name only a handful. As Fortune and 

Heffernan (2019) note, some of these subconstructs are very general and correspond to a host of 

different causal variables, whereas others are more specific, thereby casting a narrower causal 

net. Additionally, different investigators provide different operational definitions for the Central 

Eight as well as these sub-constructs, which means that the meanings of the constructs are not 

stable within the field. This is a barrier to the integration of findings from psychological research 

studies aimed at determining the causes of crime. It is also an obstacle to effective 

intradisciplinary communication.   

We also encounter a number of interfield theories when we look to sociological 

approaches to criminal behavior (Dixon, Harkness, and Wegerhoff, 2019)). The multifactorial 

model of multiple perpetrator sexual offending (MMMPSO), for example, has been used as a 

basis for understanding individual, sociocultural and situational factors that may contribute to 

group-based sexual violence (Harkins and Dixon, 2010). In their paper in this volume, Dixon, 



 16 

Harkness and Wegerhoff (2019) use the MMMPSO as a basis for developing a similar 

multifactorial theoretical framework for explaining interpersonal violent crimes, including 

assault, rape and murder, which they dub the Multifactorial Model of Interpersonal Violence 

(MMIV). On this model, violent crimes are explained by appeal to complex interactions among 

factors or features in (a) the sociocultural contexts in which individuals exist (e.g., racial or 

gender inequality), (b) the situational contexts in which they find themselves (e.g., in wartime or 

in prison; as a member of a particular family, subculture or anti-social peer group) and (c) 

psychological processes internal to the individuals themselves (e.g., identification with a 

particular social group, desire for inclusion within a social group (e.g., a gang)). Each of these 

factors working in isolation or in combination could result in an individual committing a violent 

crime and “researchers and practitioners are therefore able to draw upon this framework to 

facilitate the identification of a broad range of potential causal factors across multiple 

explanatory levels” (Dixon, Harkness, and Wegerhoff, 2019).  

The MMIV framework that Dixon and colleagues propose is similar to that of the GPSCL 

insofar as it posits a number of different constructs to pick out psychological features of persons, 

features of social groups of which they are a part and features of the ecological contexts in which 

they live. The framework is also associated with a number of risk factors for committing 

interpersonal violent crime including constructs such as: need to belong to a social group, having 

deviant peers, peer acceptance and poor parental supervision. It thus leaves open the possibility 

that different investigators will disagree about which constructs are important or interpret and 

operationally define the available constructs differently, which would clearly be a barrier to 

integrating results across different research studies aimed at explaining interpersonal violence. 

There is also some degree of overlap between the constructs associated with the MMIV model 



 17 

and those associated with the GPCSL framework, prompting a question about how, if at all, 

models in different fields of criminology may be related.  

If we instead look to experimental biology and neuroscience (Raine and Ling, 2019), we 

begin to see explanations that appear a bit more mechanistic and more multi-level—that explain 

criminal behaviors or crime-related behaviors in terms of the entities and activities situated at 

different levels of organization within those human beings who commit crimes. For example, 

subcortical areas of the brain, including the brainstem, amygdala, medial prefrontal cortex, 

parahippocampus and visual cortex have been described as being involved in subconscious 

threat detection, a construct that is operationally defined in terms of the activation of neural 

circuits taken to comprise an “innate alarm system”. This system is understood to be an 

evolutionary adaptation; individuals who had this system would have been able to avoid threats 

to their survival and thus their reproductive success in the evolutionary history of the species. 

Evidence to date suggests that in persons with post-traumatic stress disorder (PTSD), this innate 

alarm system is “hyperreactive” in response to “fear and trauma related stimuli” (Lanius et al., 

2017, p. 109). Thus, when persons with PTSD experience a threat in the form of such stimuli, 

they may react violently, resulting in them committing a criminal act.  

The interesting question with respect to this mechanistic explanation and similar 

explanations in the biological sciences that aim to link psychological traits and functions to 

structural features of organisms, is whether the constructs investigators use, like subconscious 

threat detection, fear and trauma are operationally defined across investigators in a similar way. 

Sullivan (2009, 2016) has argued previously that because investigators in neuroscience are at 

liberty, just as investigators in other areas of science, to define constructs generally and 



 18 

operationally as they deem appropriate, this poses a challenge for integrating results from the 

same and different areas of neuroscience into a common explanation of a common phenomenon.  

Although I have considered only a very small subset of the kinds of theories, models and 

explanations that we find in contemporary criminology, if we view them from the perspective of 

the conceptual tools identified in the previous section, some general themes emerge. First, 

criminology spans multiple levels of analysis and nowhere do we encounter instances of 

relationships between different fields that conform to Nagel’s model of intertheoretic reduction. 

Rather, when investigators working within a field or across fields recognize the limitations of a 

specific disciplinary approach for explaining criminal behavior, they look to conceptual and 

theoretical resources in other fields for guidance. Across those fields that comprise contemporary 

criminology, this has resulted in a diverse array of interfield theories and mechanistic 

explanations that are used to understand criminal behavior.  

Second, as a general rule, it does not appear that investigators from within the same or 

across different fields are actively collaborating currently to solve explanatory problems of 

mutual interest. Rather, researchers from different fields import a variety of different conceptual, 

theoretical, or explanatory tools from other fields to elaborate their theoretical models or 

mechanistic explanations. Indeed, they are at liberty to include those causal variables (i.e., 

constructs) within their models that they consider most relevant and emphasize the ones they 

consider to be most important. We see this in all three of the examples that I considered. For 

example, biological variables play less of a role in the GPCSL than psychological variables, and 

sociological variables are also not as prominent as psychological ones. As Fortune and Heffernan 

(2019) note, revisions to the model have led to sociological variables playing a less prominent 

role in the model over time. Similarly, sociological variables are prominent in the MMIV 



 19 

framework, psychological variables are far less so, and biological variables play no obvious role. 

The mechanistic explanation of criminal behavior refers to psychological constructs (“fear”) but 

primarily includes a description of entities and activities situated at different levels of 

organization in the brain. As McGee and Farrington (2019) put it, differences in emphasis on 

causal variables across theories tend to correlate with the orientation of the researchers who have 

developed these theories. It seems reasonable to expect that an investigator’s training will impact 

the importance they place on specific causal variables due to better familiarity with those 

variables.  

Under ideal circumstances in which investigators within each field of criminology (e.g., 

psychology, sociology) (1) agreed on how to define theoretical constructs used within that field 

to pick out causal variables and explanatory targets and (2) collectively engaged in trying to 

ensure the validity of those constructs, we might expect that developing intrafield theories of 

mechanistic explanations would be straightforward. Yet, we do not encounter such ideal 

circumstances in the fields that comprise contemporary criminology. Instead we find that 

investigators working within and across fields, even those who use the same theoretical 

framework, may put forward different general and operational definitions for the same constructs 

(Fortune and Heffernan, 2019; McGee and Farrington, 2019; Ward and Carter, 2019), barring the 

possibility of seamlessly integrating research findings from different studies purportedly related 

to the same construct. Additionally, the constructs used to pick out explanatory targets and the 

causal variables used to explain them are in some cases “lumpy”—grouping together 

heterogenous as opposed to homogeneous phenomena or grouping together multiple variables 

that ought to be treated of independently (Fortune and Heffernan, 2019; Ward and Carter, 2019; 

Ward and Fortune, 2016; Ward, Wilshire, and Jackson, 2018).  



 20 

What we find, then, when we look across the conceptual, explanatory and theoretical 

landscape of contemporary criminology is unbridled and uncoordinated intradisciplinary and 

interdisciplinary pluralism. Investigators within and across the various fields represented are at 

liberty to define terms as they deem appropriate, combine insights from whatever areas of 

science they deem relevant to their investigative and explanatory aims, and use whatever 

methods they regard as adequate for achieving their goals. Such conceptual, theoretical and 

methodological pluralism is an impediment to the development of integrative explanations of 

criminal behavior.  

 

The way forward 

Despite the absence of conceptual and methodological coordination within and across 

those fields that comprise contemporary criminology, investigators do share at least one goal in 

common—they want to identify the causes of criminal behavior for the purposes of explanation, 

prediction and control. This goal is implicit in the two of the models of scientific progress I 

considered above. Intertheoretic reduction is consistent with the idea that the ultimate aim of 

science is to explain phenomena by reference to interactions among the most fundamental units 

of matter. Mechanistic explanation, in contrast, recognizes that mechanisms are multi-level yet is 

associated with a view of progress in which phenomena (behaviors of mechanisms) are 

explained only at that point at which our explanations track the causal structure of the world. On 

both models of scientific progress, science is supposed to move towards the etiological and track 

the real-world causes of complex phenomena like crime. The problem is that neither model 

explains precisely how to coordinate practices within and across different areas of science to 

achieve this goal.  



 21 

We find at least two suggestions in the philosophy of science literature for how fields that 

investigate complex phenomena that are struggling to make progress in the ways just described 

might achieve it. Consider a field like psychiatry, which is not so dissimilar from criminology (in 

fact, there is a lot of overlap. In the 1950’s, the American Psychopathological Association (now 

the American Psychiatric Association (APA)) was struggling to make progress in understanding 

the causes of mental illness because research scientists and practitioners approached the 

scientific study of mental illness from different vantage points. At that time, the Diagnostic and 

Statistical Manual of Mental Disorders (DSM-1) contained many psychodynamic terms like 

“conversion”, and “mental conflict”—constructs that were used in different ways by practitioners 

coming from different theoretical backgrounds (Hempel, 1959/1965; Haslam, 2013). This 

prompted miscommunication and was regarded as a problem that needed to be solved if 

psychopathology was to make progress in understanding mental illness.  

In an invited address to the APA (1959/1965), philosopher of science Carl Hempel offered 

one solution. He suggested that investigators having different theoretical backgrounds ought to 

begin by agreeing what the kinds in the classification system should be. Then they should specify 

the criteria for individuating those kinds in terms of observable surface features. The resulting 

operationally defined categories were intended to facilitate communication among investigators 

having different theoretical perspectives who were engaged in a common research project but 

lacked a shared vocabulary. Operational definitions, however, were not the endgame of science 

according to Hempel; they served only as an important practical starting point. Sciences should 

strive to move away from operational definitions to conceptual taxonomies having systematic or 

theoretical import. In other words, they should aim to discover general laws or theoretical 



 22 

principles which reflect uniformity in the subject matter under study and which provide a basis 

for explanation, prediction, scientific understanding, and control.  

One problem with Hempel’s solution is the lack of direction he provides as to how to ensure 

a workable and stable taxonomy that facilitates mutual understanding across scientists who share 

an explanatory goal in common, when this same taxonomy is supposed to be modified in light of 

causal discovery. If taxonomic categories are meant to be revised as new information about 

mechanisms comes in, the categories will not be stable. Yet if the categories are in flux until 

science “gets the worlds right”, it is not clear how they will continue to facilitate the 

interdisciplinary communication for which they were originally intended.  

In their contributions to this volume, McGee and Farrington (2019) and Fortune and 

Heffernan (2019), though they are working in different fields of criminology, make suggestions 

for resolving construct instability within their respective fields that resemble Hempel’s proposal 

to the APA. Specifically, they claim that an important pathway to progress is for investigators 

working within a given field to collectively identify a set of intradisciplinary constructs and 

collaboratively put forward definitions of those constructs with an eye towards developing a 

classification system that will track the causes of crime and allow for findings related to the 

phenomena of interest to be integrated across research studies. To these proposals, I will only 

add that such aims will not be attainable if there fails to be ongoing collaboration among 

investigators working in each of these fields—a kind of collaboration that is not currently 

present. Moreover, if the ultimate aim of stabilizing the constructs in these fields is the 

integration of findings across different fields that investigate criminal behavior (i.e., 

interdisciplinary integration), then such collaborative endeavors ought to be interdisciplinary. 

Fortune and Heffernan favor the development of field-specific or what they call “local theories” 



 23 

first, once investigators within a given field such as forensic psychology agree on a classification 

scheme. However, if multi-level explanations are indeed the goal, conceptual, theoretical and 

explanatory practices eventually will have to be coordinated within and across different scientific 

fields (See Sullivan 2018 for further discussion about strategies for coordinating practice across 

fields).  

An alternative to Hempel’s proposal recently proposed in the philosophical literature 

advocates a different route for moving from the descriptive to the etiological. Instead of 

investigators from different areas of science collaboratively operationally defining their 

categories, they can begin looking at the wealth of data collected across different areas of science 

as a means to determine if any “robust patterns” emerge (Tabb and Schaffner, 2017). The 

patterns that will be detected will always be perspective-relative, though proponents of this 

alternative insist that this does not compromise their reality. Patterns exist in the world and while 

“there are many possible grounds for delineating [them] [. . ] on the basis of our diverse 

theoretical frameworks, [. . .] only some will be meaningful” (Tabb and Schaffner, 2017, 344). 

This view is consistent with the idea that some theoretical vantage points may be more 

advantageous than others for detecting meaningful patterns. Moreover, only important patterns 

ought to serve as a basis for deciding on the categories that are to become the focus of 

interdisciplinary inquiry. This of course, however, leaves open the possibility that researchers 

hailing from different theoretical backgrounds may disagree on which perspective(s) should take 

priority.  

In two of the papers contained in this volume (Durrant, 2019; Ward and Carter, 2019; See 

also Durrant and Ward, 2015), we encounter a prescription for future progress in criminology 

that exemplifies the “robust patterns” approach. In their paper, Ward and Carter (2019, pp) 



 24 

identify a number of problems with current systems that classify criminals on the basis of the 

offenses they commit (e.g., sexual or violent crime), the likelihood that they will commit an 

offense (“risk level”), the number and type of dynamic risk factors they present with (where the 

more factors that are present, the more at risk the individual is for offending or recidivism) and 

“their underlying antisocial dispositions”.  They claim that these classification systems are not 

explanatory nor do they point to causal variables that could be targeted to prevent criminal 

behavior. Moreover, the behavioral patterns to which they point are not sufficiently “robust”—

they do not detect meaningful categories that suit either pragmatic or explanatory interests.  

Ward and Carter (2019) advocate a classification strategy that begins with the recognition 

that a set of robust “behavioral patterns” across the phenomenal landscape of crime emerge when 

one adopts a different theoretical perspective – namely, when one asks: What is the function of 

an individual’s criminal behavior? and What goal are they trying to achieve by engaging in that 

behavior? The basic idea is to determine what function committing a crime serves for the 

individual. As they put it, from this “functional perspective”, which has its orgins in evolutionary 

biology (Durrant and Ward, 2015), “criminal behavior is remarkably like normal behavior and 

may well have adaptive outcomes for the person concerned, despite being necessarily 

undesirable from a social/legal perspective” (Ward and Carter, (2019) pp. ). Shifting to this 

perspective, they claim puts the focus on interactions between “core motivational systems” and 

“contextual variables” in an individual’s environment. Moreover, it integrates data from 

cognitive neuroscience, evolutionary biology, psychopathology, disability research and ethology, 

to name only a handful of fields.  

Ward and Carter’s proposal bears some resemblance to a recent initiative instituted by the 

US National Institute of Mental Health (NIMH), the Research Domain Criteria Project 



 25 

(Cuthbert and Insel, 2013). Proponents of RDoC believe that current categories of mental illness 

are insufficient for guiding the discovery of causes and that mental disorders are better 

understood as disruptions in different domains of psychological and behavioral functioning. To 

this end, the RDoC task forces identified a set of constructs designating psychological and 

behavioral functions that are intended to facilitate the integration of data from various scientific 

fields including psychology, systems neuroscience, and neurobiology. The success of RDoC, 

however, is taken to be contingent on the maintenance of an organizational infrastructure that 

ensures the collective stabilization of constructs designating psychological and behavior 

functions so as to facilitate the development of integrative mechanistic explanations (See also 

Kutschenko, 2011; Ankeny and Leonelli, 2015, 2016; Sullivan 2016, 2017; Tabb, 2017). Thus, 

we might imagine that the development of a similar interdisciplinary infrastructure will be 

required if the shift to a functional perspective in criminology is to be successful.  

 
 
Conclusion  
 
Contemporary criminologists agree that understanding, explaining and preventing crime requires 

an interdisciplinary approach. As the papers in this special issue indicate, the conceptual, 

methodological, theoretical and explanatory pluralism that we encounter within and across the 

scientific fields that comprise contemporary criminology are barriers to understanding and 

explaining crime. Overcoming these obstacles will require intensive collaborative efforts within 

and across these different fields to ensure conceptual clarity and methodological and explanatory 

rigor. The meta-analytic work showcased in this volume is an important first step for facilitating 

intra- and inter-disciplinary dialogue requisite for progress.  

 
 



 26 

Bibliography 

Andrews, D., Bonta, J. and Wormith, S. (2011). The Risk-Need-Responsivity (RNR) Model 

Does Adding the Good Lives Model Contribute to Effective Crime Prevention? Criminal 

Justice and Behavior, 38(7), 735-755.  

Ankeny, R. and Leonelli, S. (2016). Repertoires: A Post-Kuhnian Perspective on Scientific 

Change and Collaborative Research.” Studies in History and Philosophy of Science, 60, 18-

28. 

Ankeny, R. and Leonelli, S. (2015). Repertoires: How to Transform a Project into a Research 

Community. Bioscience, 65(7), 701-708.  

Baron, E. and Sullivan, J. (2018). Judging Mechanistic Neuroscience: A Preliminary Conceptual 

Analytic Framework for Evaluating Scientific Evidence in the Courtroom. Psychology, 

Crime and Law, 24(3), 334-351. 

Beattie, J. (1959). Understanding and Explanation in Social Anthropology. British Journal of 

Sociology, 10, 45-60.  

Bechtel, W. (2008). Mental Mechanisms: Philosophical Perspectives on Cognitive 

Neuroscience. New York: Taylor & Francis.  

Bechtel, W. (1986). Integrating Scientific Disciplines: Case Studies from the Life Sciences. 

Hingham, MA: Kluwer Academic Publishers.  

Bechtel, W. and Richardson, R. (1993). Discovering Complexity: Decomposition and 

Localization as Strategies in Scientific Research. Princeton: Princeton University Press.  

Burns, J. and Swerdlow, R. (2003). Right orbitofrontal tumor with pedophilia symptom and 

constructional apraxia sign. Archives of Neurology, 60(3), 437-440.  

Chang, H. (2010). Operationalism. Stanford Encyclopedia of Philosophy (online). 



 27 

Churchland, P. and Sejnowski, T. (1988). Perspectives on Cognitive Neuroscience. Science, 

242(4879), 741-745.  

Cronbach, L. and Meehl, P. (1955). Construct Validity in Psychological Tests. Psychological 

Bulletin, 52, 281-302. 

Cummins, R. (1983). The Nature of Psychological Explanation. Cambridge, MA: MIT Press. 

Cummins, R. (1975). Functional analysis. Journal of Philosophy, 72, 741-65.  

Craver, C. (2007). Explaining the Brain: Mechanisms and the Mosaic Unity of Neuroscience. 

Oxford: Oxford University Press. 

Darden, L. & Maull, N. (1977). Interfield Theories. Philosophy of Science, 44(1), 43-64. 

Day, A., Tamatea, . and Gaia, . (2019). Scientific Inquiry and Offender Rehabilitation: The 

Importance of Epistemic and Prudential Values. Psychology, Crime and Law, (this 

volume). 

Dixon, . Harkness, L., and Wegerhoff, . (2019). Incorporating Sociocultural and Situational 

Factors into Explanations of Interpersonal Violent Crime. Psychology, Crime and Law, 

(this volume). 

Durrant, R. (2019). Evolutionary Approaches to Understanding Crime: Explaining the Gender 

Gap in Offending. Psychology, Crime and Law, (this volume).  

Durrant, R. and Ward, T. (2015). Evolutionary Criminology: Towards a Comprehensive 

Explanation of Crime. Amsterdam: Academic Press.  

Fodor, J. (1968). Psychological Explanation: An Introduction to the Philosophy of Psychology. 

New York: Random House.  

Fortune, C.A. and Heffernan, . (2019). The Psychology of Criminal Conduct. Psychology, Crime 

and Law, (this volume).  



 28 

Giere, R. (2010). Scientific Perspectivism. Chicago: University of Chicago Press. 

Haig, B. (2019). The Importance of Scientific Method for Psychological Science. Psychology, 

Crime and Law, (this volume).  

Haslam, N. (2013). Reliability, Validity, and the Mixed Blessings of Operationalism. In K.W.M. 

Fulford, M. Davies, R. Gipps, G. Graham, J. Sadler, G. Tanghellini, and T. Thornton, 

(Eds), The Oxford Handbook of Philosophy and Psychiatry. Oxford: Oxford University 

Press.  

Hempel, C. & P. Oppenheim. (1948). Studies in the Logic of Explanation. Philosophy of 

Science, 15, 135-75.  

Kutschenko, L. (2011). How to Make Sense of Broadly Applied Medical Classification Systems: 

Introducing Epistemic Hubs. History and Philosophy of the Life Sciences, 33(4), 583-601.  

Machamer, P., Darden, L. and Craver, C. (2000). Thinking about Mechanisms. Philosophy of 

Science, 67(1), 1-25.  

Machamer, P. and Sullivan, J. (2001). Leveling reduction. Philsciarchive.  

McGee, T. and Farrington, D. (2019). Developmental and Life-Course Explanations of Crime. 

Psychology, Crime and Law, (this volume).  

Mitchell, S. (2003). Biological Complexity and Integrative Pluralism. Cambridge: Cambridge 

Studies in Philosophy and Biology.   

Nagel, E. (1961). The Structure of Science: Problems in the Logic of Scientific Explanation. 

New York: Harcourt, Brace & World. 

Oppenheim, P. and Putnam, H. (1958). Unity of Science as a Working Hypothesis, Minnesota 

Studies in the Philosophy of Science, 2, 3-36.  

People v Goldstein, 14 AD (3d) 32 (NY App Div 2005).  



 29 

Piccinini, G. and Craver, C. (2011). Integrating Psychology and Neuroscience: Functional 

Analysis as Mechanism Sketches. Synthese, 183(3), 283–311. 

Potochnik, A. (2010). Levels of explanation reconceived. Philosophy of Science, 77(1), 59-72.  

Potochnik, A. and McGill, B. (2012). The limitations of hierarchical organization. Philosophy of 

Science, 79(1), 120-140.  

Raine, A. and Ling, . (2019). Biological explanations of Criminal Behavior. Psychology, Crime 

and Law (this volume).   

Rafter, N. (2011). Origins of Criminology. In M. Bosworth and C. Hoyle (Eds.), What is 

Criminology? Oxford: Oxford Scholarship Online.  

Salmon, W. (1989). Four Decades of Scientific Explanation. Minneapolis: University of 

Minnesota Press. 

Shapere, D. (1974). Scientific Theories and Their Domains. In F. Suppe, (Ed.), The Structure of 

Scientific Theories. Urbana: Univesrity of Illinois Press, pp. 518-565. 

Stinson, C. (2015). Mechanisms in psychology: Ripping Nature at Its Seams. Synthese, 193(5), 

1585-1614.  

Stinson, C. and Sullivan, J. (2017). Mechanistic Explanation in Neuroscience. In S. Glennan and 

P. Illari, (Eds.), The Routledge Handbook of Mechanisms and Mechanical Philosophy. 

London: Routledge: 375-388.  

Sullivan, J. (2017). Coordinated Pluralism as a Means to Facilitate Integrative Taxonomies of 

Cognition. Philosophical Explorations, 20(2), 129-145.  

Sullivan, J. (2016). “Construct Stabilization and the Unity of the Mind-Brain Sciences”, 

Philosophy of Science, 83(5), 662-73.  



 30 

Sullivan, J. (2015). Neuroscientific Kinds Through the Lens of Scientific Practice. In C. Kendig 

(Ed.), Natural Kinds and Classification in Scientific Practice. London: Routledge.  

Sullivan, J. (2014). “Is the Next Frontier in Neuroscience A Decade of the Mind? In Charles 

Wolfe (Ed.), Brain Theory: Essays in Critical Neurophilosophy (pp. 45-67). New York: 

Palgrave-Macmillan.  

Sullivan, J. (2010). Reconsidering Spatial Memory and the Morris Water Maze. Synthese, 

177(2), 2661-283. 

Sullivan, J. (2009). The Multiplicity of Experimental Protocols: A challenge to Reductionist and 

Nonreductionist Models of the Unity of Science. Synthese, 167, 511-39.  

Tabb, K. (2017). Philosophy of Psychiatry After Diagnostic Kinds. Synthese. 

https://doi.org/10.1007/s11229-017-1659-6 

Tabb, K. and Schaffner, K. (2017). Causal pathways, random walks, and tortuous paths: Moving 

from the descriptive to the etiological in psychiatry. In K. S. Kendler and J. Parnas, (Eds.) 

Philosophical Issues in Psychiatry IV (pp. 342-360). Oxford: Oxford University Press. 

Ward, T. and Carter, A. (2019). The Classification of Offending and Crime Related Problems: A 

Functional Perspective. Psychology, Crime and Law (this volume).  

Ward, T., Polaschek, D. and Beech, A. (2005). Theories in Sexual Offending. Wiley.  

Ward, T., Wilshire, C. & Jackson, L. (2018). The Contribution of Neuroscience to Forensic 

Explanation, Psychology, Crime & Law, 24(3), 195-209.  

Weaver, B. (2019). Understanding Desistance: A Critical Review of Theories of Desistance. 

Psychology, Crime and Law, (this volume).