A person repeatedly checks their phone after sending an important message. One explanation might describe the behaviour psychologically: they are uncertain about how the message will be received and are seeking reassurance. A behavioural explanation might point to the fact that checking has occasionally been rewarded by receiving the anticipated reply, making repetition more likely. A relational explanation might note that delayed responses have previously carried interpersonal consequences. A physiological account might examine arousal, attention and the systems involved in detecting uncertainty or potential threat. An environmental explanation might add that smartphones make checking almost effortless by keeping socially relevant information continuously available.
These explanations need not be rivals. They may be describing different aspects of the same event, answering different questions about how the behaviour arose, why it persists, what conditions make it more likely, or through what processes it is implemented. The mistake begins when one explanation is treated as though its validity automatically makes the others unnecessary.
This is a recurring problem in discussions of human behaviour. Psychological explanations are sometimes dismissed as superficial once a neural correlate has been identified. Biological accounts are sometimes treated as inherently more scientific because they describe processes at a smaller physical scale. Social explanations can be presented as alternatives to individual ones, while environmental conditions are relegated to the status of background context rather than recognised as part of the causal structure of behaviour. At other times the opposite mistake occurs: biological findings are dismissed because they do not capture lived experience, as though the usefulness of one explanatory level required rejecting another.
Philosophy of science provides good reasons to resist these moves. Different levels of analysis can make non-redundant contributions to explanation, and progress often depends on understanding how those contributions relate rather than deciding prematurely which level is the “real” one. Contemporary work in psychology has renewed this argument, drawing on traditions including Tinbergen’s questions in biology and Marr’s levels of analysis in cognitive science to show how apparently competing explanations can instead address distinct explanatory problems.
The result is not the vague claim that everything affects everything. A serious multilevel account requires much more discipline than that. It requires asking what exactly is being explained, which question each explanation answers, what evidence supports the proposed relationship, and where the limits of the explanation lie.
Explanation begins with the question being asked
The phrase an explanation of behaviour can give the impression that there should be one complete answer waiting to be discovered. In practice, explanation is always relative to a question, even when the question remains implicit. Asking why a person avoided a social event is not yet specific enough to determine what kind of answer would count as satisfactory.
One person may be asking what the individual expected would happen if they attended. Another may want to know how avoidance became more frequent over time. A clinician might be interested in whether a recognised psychological condition contributes to the pattern. A neuroscientist might investigate the neural processes involved in threat evaluation. A sociologist might ask how social exclusion, economic position or cultural expectations alter participation. Each inquiry concerns the same broad phenomenon while selecting a different feature of it for explanation.
This multilevel way of approaching explanation also connects with the broader account of human functioning developed in What Do We Mean by Human Organisation?, where the emphasis is on relations between components, conditions and contexts rather than treating any single variable as sufficient on its own. Human behaviour can be examined at several levels without assuming that one level automatically cancels or replaces the others.
This is not merely a matter of looking at the same object from several subjective perspectives. Different questions can require genuinely different evidence. Evidence about neural activation cannot by itself establish what social consequences a person anticipated. Evidence that someone expected rejection does not identify the neural processes through which that expectation was implemented. A description of cultural norms may explain why particular outcomes carry significance without explaining why two people exposed to similar norms respond differently.
The philosopher Sam Wilkinson makes a related distinction between levels of explanation and kinds of explanation, emphasising that different explanatory concerns generate different kinds of legitimate “why” questions. This is a useful corrective to the assumption that explanation is simply a competition in which increasingly microscopic accounts progressively replace broader ones.
A good explanation therefore begins by specifying its target. Before asking whether an explanation is adequate, it is worth asking: adequate for explaining what?
Description is not yet explanation
Another difficulty appears because descriptions can sound explanatory without actually establishing why something happened. Saying that a person avoided a conversation because they were anxious may provide useful information, but depending on how anxious is being used, the explanation can become circular. If anxiety is inferred primarily from the fact that the person avoided the conversation, then saying that avoidance occurred because of anxiety may simply redescribe the behaviour in psychological language.
The same problem can arise with biological terminology. Saying that a behaviour occurred “because of dopamine”, “because of cortisol”, or “because the nervous system was activated” can create an impression of causal depth while leaving the relevant mechanism unspecified. The presence of a physiological correlate does not by itself establish that the correlate caused the behaviour, nor does naming a biological process automatically explain why that process occurred under those particular conditions.
Research literacy requires distinguishing observation, association, mechanism and causal inference. A study may show that two variables covary without demonstrating which produces the other. An experimental manipulation may provide stronger evidence of causal influence but still operate under limited conditions. A mechanism may be plausible without yet being demonstrated in the population or context under discussion.
This matters especially in public communication, where technical vocabulary can acquire explanatory authority simply because it sounds mechanistic. A biological description is not necessarily a deeper explanation merely because it refers to smaller physical components. The scientific value of an explanation depends on what it accounts for, how well its claims are supported and whether it adds information that was not already contained in the phenomenon being described.
Lower levels are not automatically deeper levels
There is an intuitive hierarchy in much popular science. Social explanations sit near the surface, psychological explanations lie underneath them, and eventually neuroscience, genetics or molecular biology reveal what is reallyhappening. According to this picture, moving downwards through physical scales means moving closer to the true cause.
Scientific practice is considerably less tidy.
Reductionism has played an enormously productive role in science because complex systems can often be investigated by examining their components. Neuroscience, molecular biology and genetics have all advanced by identifying mechanisms that would have remained invisible at broader scales. The problem arises when methodological reduction — studying parts in order to understand a system — becomes explanatory reductionism, where findings at one level are assumed to make explanations at other levels dispensable.
Debates about neurobiological reduction illustrate why that move is controversial. Neural and molecular processes undoubtedly matter to behaviour, but explaining those processes does not necessarily exhaust what needs to be explained about an organism acting within an environment. Reviews of neurobiological reduction have therefore argued against both extremes: neither treating molecular and cellular detail as the only legitimate form of explanation nor dismissing such detail as irrelevant to higher-level phenomena.
Consider reading a sentence. The activity depends on physical processes in the nervous system, but an explanation of why a reader interprets a particular word as sarcastic may require information about language, previous sentences, social conventions and the relationship between speaker and listener. Describing neuronal activity with perfect precision would not automatically supply those facts.
This does not make the psychological or social explanation somehow non-physical. It means that explanatory usefulness depends partly on the organisation and scale of the phenomenon being investigated. The fact that every psychological event depends on physical processes does not imply that every useful psychological explanation can be replaced by a description of those processes.
Marr’s levels illustrate why explanations can be complementary
David Marr’s influential framework in cognitive science offers a particularly clear example. Marr distinguished questions about what a system is doing and why that task matters, questions about the representations and procedures through which the task is performed, and questions about how those procedures are physically implemented.
These levels were developed in relation to information-processing systems, particularly vision, rather than as a universal model of all human behaviour. Their broader importance lies in demonstrating that distinct explanatory questions can make non-redundant contributions to understanding the same system. Bechtel and Shagrir argue that Marr’s three levels provide different perspectives required for understanding information-processing mechanisms, with each level contributing information that the others do not simply duplicate.
Suppose a person is navigating an unfamiliar city. We could ask what problem their navigational system is solving, how spatial information is represented and updated, and how the relevant processes are physically implemented in the nervous system. Discovering the neural implementation would be scientifically valuable, but it would not make questions about the computational problem or representational strategy meaningless.
The broader lesson is not that every human behaviour must be forced into Marr’s three categories. It is that explanation can be plural without being incoherent. Several explanations can concern the same phenomenon while differing in what they are trying to explain.
Tinbergen showed that even “why” contains several questions
A similar point emerged from biology through Nikolaas Tinbergen’s famous four questions concerning behaviour. Tinbergen distinguished questions relating broadly to mechanism, development, function and evolutionary history. Later assessments of the framework have argued that these questions remain valuable because their answers provide complementary rather than mutually exclusive forms of understanding.
Consider an animal avoiding a predator. We could investigate the immediate sensory and physiological mechanisms that generate escape behaviour. We could study how the behaviour develops across the animal’s lifetime. We could ask what consequences the behaviour has for survival and reproduction, or investigate how the behaviour emerged across evolutionary history.
An answer to one does not automatically answer the others. Identifying the immediate neural mechanism of escape does not tell us how the response develops. Demonstrating the survival value of avoidance does not specify the physiological machinery through which avoidance occurs.
Again, human behaviour cannot simply be reduced to Tinbergen’s framework. Human social, cultural, linguistic and institutional environments introduce additional explanatory dimensions. The importance of the example is methodological: even within biology, asking why does this behaviour occur? turns out to contain several distinct scientific questions.
Psychological explanations are real explanations
Psychological explanations are sometimes treated as provisional placeholders that neuroscience will eventually replace. This view assumes that concepts such as expectation, attention, belief, memory or decision-making are useful only until their underlying biological mechanisms are fully known.
That assumption is difficult to sustain in practice. Psychological variables often describe relations and functions that are relevant precisely at the level of the behaving organism. If someone avoids an activity because they expect humiliation, the expectation can be causally relevant regardless of the fact that the expectation is physically realised in a nervous system. Knowing more about its neural implementation can enrich the explanation without necessarily removing the need to know what was expected.
There is also an important distinction between the existence of a lower-level implementation and the adequacy of a lower-level explanation. Every digital document is physically instantiated in hardware, yet explaining why a sentence contains a grammatical error rarely requires discussion of transistor states. The physical implementation is necessary for the document to exist, but it is not necessarily the explanatory level relevant to the question being asked.
Psychology is not identical to software engineering, and analogies between minds and computers have important limitations. The example nevertheless illustrates a general point: dependence does not entail explanatory replacement. One level can depend on another while retaining explanatory questions that are most naturally formulated at its own scale.
Behavioural explanations add a different kind of information
Behavioural analysis directs attention towards what people do, the conditions under which behaviour occurs and the consequences that may help maintain it. This can reveal patterns that are obscured when explanation concentrates exclusively on internal states.
Suppose someone repeatedly seeks reassurance whenever they become uncertain about a decision. A psychological description might identify fear of making a mistake. A behavioural account could examine what happens after reassurance is sought. If uncertainty decreases temporarily, reassurance-seeking may become increasingly likely the next time uncertainty appears.
The two explanations need not compete. One concerns what the person anticipates or experiences; the other concerns a pattern between behaviour and its consequences. They can intersect without becoming identical.
Behavioural explanations are also useful because they encourage attention to repeated observations rather than relying entirely on retrospective narratives. Someone may sincerely believe that they seek reassurance only when decisions are especially difficult, while closer observation shows that reassurance occurs particularly when another person expresses disagreement. That pattern adds information which the person’s original explanation did not contain.
The point is not that behaviour provides a more objective or fundamentally superior level. Behavioural descriptions also require interpretation, operational definitions and decisions about what variables matter. Their value lies in revealing a different structure of evidence.
Relationships can be explanatory, not merely contextual
Individual-level explanations often treat relationships as environments within which psychological processes happen. Yet relationships can themselves contain organised patterns that influence behaviour in ways that cannot be captured adequately by describing either participant in isolation.
A person may speak confidently in most settings but become highly hesitant in one particular relationship. Describing them globally as lacking confidence would miss the variation. An explanation focused only on personality might similarly obscure the fact that hesitation emerges specifically in response to predictable patterns of criticism, interruption or withdrawal.
The relationship changes the conditions under which capacities are expressed. This does not mean that the relationship alone “causes” the person’s behaviour, because prior learning, expectations, temperament and current circumstances may also contribute. It means that explanatory adequacy sometimes requires analysing interactions rather than locating the cause exclusively within one individual.
This becomes especially important when concepts such as resilience, self-esteem or emotional regulation are used to explain difficulties arising in harmful environments. If an environment systematically punishes particular behaviours, telling the individual to become more confident may misidentify the relevant causal structure. The question is no longer only what capacity the person possesses, but what happens when that capacity is expressed within a particular relational system.
Environments do not merely trigger pre-existing behaviour
Environmental explanations are sometimes treated as descriptions of triggers acting on an otherwise self-contained person. This framing understates the extent to which environments participate in shaping what behaviour becomes possible, useful or costly.
A workplace with unpredictable sanctions may produce different patterns of communication from one in which disagreement is tolerated. A smartphone environment in which social information is available continuously makes checking behaviour easier to perform and easier to repeat. Financial insecurity changes the consequences associated with risk. Physical accessibility changes which actions are realistically available, while cultural expectations influence the social meaning attached to them.
These conditions do not determine behaviour completely. People exposed to similar environments can behave differently because their histories, resources, expectations and capacities differ. Environmental explanation is therefore not an alternative determinism in which context replaces psychology as the single cause.
The more useful point is that behaviour emerges through relations between organisms and environments. Asking only what is happening “inside” the person can obscure conditions without which the behaviour would look very different.
Physiological explanations matter without becoming universal explanations
Human behaviour is embodied, and physiological processes place genuine constraints on what people can do. Sleep deprivation alters attention and decision-making. Pain can change behaviour. Hormonal, autonomic and neural processes contribute to perception, action, motivation and emotion. Ignoring physiology would therefore produce an incomplete account of many phenomena.
The problem appears when physiological explanation is given a privileged status regardless of the question being asked. A measurable bodily process can feel more scientifically secure than a relational or psychological variable because it appears concrete. Yet measurement precision does not guarantee explanatory relevance.
A precisely measured biological variable may explain very little about why a behaviour occurred in one context rather than another. Conversely, a relatively difficult-to-measure social variable may be indispensable to understanding the event. The quality of explanation cannot therefore be ranked simply according to how physically small or technologically measurable its variables are.
This point is particularly important in contemporary public discourse about the brain and nervous system. Biological terminology can be useful when evidence warrants it, but invoking physiology should increase explanatory precision rather than merely confer scientific authority on a claim.
One phenomenon can have causes at several scales
The possibility of multiple explanatory levels becomes clearer when we move away from the idea that an event must have one cause.
Suppose a person sleeps poorly before an important professional presentation. At a physiological level, arousal may interfere with the processes required for sleep. Psychologically, the person may be anticipating failure or social evaluation. Behaviourally, they may repeatedly rehearse the presentation late into the night. Environmentally, their phone may continue delivering work-related messages. Organisationally, the workplace may attach unusually high consequences to the presentation.
These factors can interact. Anticipation may increase arousal, which makes sleep more difficult. Difficulty sleeping may create more opportunity for rehearsal. Rehearsal may maintain attention on the presentation, while messages from colleagues reinforce its perceived importance. None of these observations requires the claim that every factor contributes equally.
These interactions also matter over time. When several contributing conditions change at different rates, the resulting pattern of behaviour may accelerate, stall or appear to reverse rather than changing smoothly. This temporal dimension is examined more directly in Why Change Can Be Nonlinear, where the focus is on why meaningful change need not follow a steady or proportional trajectory.
A strong explanation would investigate which relationships are actually supported by evidence in the particular case. A weak explanation would simply list every conceivable influence and call the result holistic.
This difference is critical. Explanatory pluralism is not an invitation to accumulate causes without discrimination. It increases rather than decreases the need for careful causal reasoning.
Multiple levels do not mean every explanation is equally good
Once we accept that more than one explanatory level can matter, another error becomes possible: treating all explanations as equally legitimate because each represents a different “perspective”.
Scientific pluralism is not relativism. Explanations can still be poorly supported, circular, vague, causally implausible or contradicted by evidence. Two explanations can genuinely compete when they make incompatible claims about the same question.
If one hypothesis predicts that a behaviour should increase under a certain condition and another predicts that it should decrease, evidence can discriminate between them. If an explanation requires an event to have preceded another event but the chronology shows the opposite, the explanation has a problem. If a proposed mechanism is inconsistent with established evidence, calling it a different explanatory level does not rescue it.
Levels become complementary when they answer different questions or describe different parts of a causal structure. They are not complementary merely because they use different terminology.
This distinction prevents “multilevel explanation” from becoming a way of protecting claims from criticism. The scientific task remains to specify what each explanation claims and what observations would count against it.
Integration is more difficult than accumulation
It is relatively easy to produce a paragraph containing psychological, neurological, social and behavioural terminology. It is much harder to demonstrate how those explanations relate.
Research on levels of analysis increasingly emphasises this problem. Simply gathering information at multiple levels does not automatically produce integration; the relationships between levels must themselves be explained. A useful multilevel account should therefore identify where one process constrains another, how variables interact across time, and which explanatory links are established rather than merely assumed.
A related distinction appears in discussions of individual change. Possessing information, insight or conceptual understanding does not necessarily mean that different aspects of functioning have become coordinated around that understanding. This distinction is explored further in Insight Is Not the Same as Integration, where integration is treated as something more demanding than simply knowing or understanding something intellectually.
For example, suppose social exclusion is associated with increased vigilance. A multilevel explanation might eventually connect social circumstances, psychological expectations, attentional processes, behavioural monitoring and physiological responses. But identifying variables at all five levels would not yet show that one caused another, or establish the direction and timing of those relationships.
Integration therefore requires empirical bridges. Which environmental changes alter the behaviour? Does changing expectation alter the response? What happens over repeated exposures? Are the physiological changes causes, consequences or both? Under what conditions does the pattern fail to occur?
Without those questions, the appearance of comprehensiveness can exceed the actual explanatory evidence.
Different levels can constrain one another
Although explanatory levels should not be collapsed, neither should they be imagined as completely independent. Findings at one level can constrain what is plausible at another.
A psychological theory that requires capacities inconsistent with known perceptual limitations would need revision. A neurological explanation that predicts behaviour contrary to reliable behavioural evidence would also face a problem. A social explanation that assumes a particular belief is universally present would be weakened if psychological research demonstrates substantial variation.
This reciprocal constraint is one of the reasons multilevel research can be valuable. The goal is not simply to preserve separate descriptions but to determine where they meet.
Bechtel and Shagrir’s discussion of Marr makes this point clearly: different levels retain distinct explanatory contributions while also standing in relations that allow discoveries at one level to inform what is possible at another. Recent work in psychology similarly argues that frameworks from biology and cognitive science can clarify explanatory goals and expose gaps that would otherwise remain hidden.
The implication is a form of disciplined pluralism. Explanatory levels are neither reducible to one another in every case nor free to contradict one another indefinitely.
Human behaviour is especially vulnerable to single-level stories
Single explanations are attractive partly because they simplify communication. “It is a trauma response”, “it is dopamine”, “it is attachment”, “it is conditioning”, “it is personality”, or “it is the environment” provides a compact account that feels complete enough to organise the phenomenon.
The difficulty is that human behaviour often reflects historical processes unfolding within current environments. A present behaviour may depend on what someone learned previously, what they expect now, what consequences are currently available, what other people are doing, what physical resources they possess, and what physiological state they are in at that moment.
This does not mean that every explanation requires a complete inventory of the human organism and its environment. Most scientific explanations are selective because explanation would become impossible if every causal condition had to be included. The question is whether the selection leaves out something necessary for the phenomenon being investigated.
A useful explanation is therefore not necessarily the longest explanation. It is the one that identifies the variables relevant to the question and supports the proposed relationships with appropriate evidence.
The same behaviour can have different explanations in different people
Another reason to resist single explanatory templates is that superficially similar behaviour can arise through different processes.
Two people may both avoid public speaking. One may expect humiliation following a previous experience of ridicule. Another may simply have little experience speaking publicly and therefore find the task unusually demanding. A third may be exhausted and avoiding an additional workload, while a fourth may work in an environment where public mistakes carry unusually severe professional consequences.
The observable behaviour is similar, but its organisation differs. Treating avoidance itself as proof of one underlying mechanism would therefore confuse behavioural similarity with explanatory identity.
The reverse also occurs. Different behaviours can serve similar functions. One person may avoid uncertainty by repeatedly asking for reassurance, another by researching excessively, and another by postponing decisions. Looking only at surface behaviour could conceal their functional similarity.
This is one reason why explanation requires more than classification. Giving a behaviour a name can help organise observations, but the category does not automatically reveal what produces or maintains the behaviour in a particular case.
Explanation should change what we predict
A useful test of an explanation is whether it helps us understand what should happen when relevant conditions change.
If reassurance-seeking is maintained partly because reassurance temporarily reduces uncertainty, then preventing or delaying reassurance should alter the sequence in a predictable way. If hesitation occurs primarily in one coercive relationship rather than across contexts, then changing the interpersonal environment should affect the behaviour. If sleep deprivation is contributing substantially to poor concentration, concentration should improve when adequate sleep is restored, assuming other relevant conditions remain reasonably stable.
Predictions need not always be experimentally tested in everyday settings, and ethical or practical constraints may make direct manipulation impossible. Nevertheless, asking what an explanation predicts helps distinguish causal proposals from stories that merely fit events retrospectively.
A narrative can sound plausible because it is constructed after the outcome is known. Scientific explanation becomes more demanding when it specifies what evidence would be expected if the account were correct and what findings would require reconsideration.
This is also why explanatory humility matters. Many human phenomena are multiply determined and difficult to manipulate experimentally. In such cases, the appropriate conclusion may be that several explanations remain plausible rather than pretending that available evidence identifies one definitive cause.
Explanation is not the same as intervention
Another important distinction concerns the relationship between understanding why something occurs and knowing how to change it. A factor can contribute causally to a problem without being the most useful target for intervention, while an effective intervention can sometimes work through mechanisms that are not fully understood.
Suppose a person’s current behaviour has been influenced by years of relational experience. That history cannot be changed. Yet current environments, expectations or behavioural routines may be modifiable. Conversely, a measurable biological correlate may be involved in the behaviour without being directly alterable or practically useful as an intervention target.
This distinction prevents an important reasoning error: assuming that the most fundamental explanation must identify the best point of intervention. Lower-level mechanisms may be causally important while higher-level changes remain more accessible. Altering a social environment can sometimes change biological processes downstream; changing behaviour can alter environmental consequences; physiological changes can affect psychological functioning.
Causal systems do not respect the simple hierarchy implied by phrases such as “root cause”. There may be several intervention points, and the most useful one depends on feasibility, evidence, risk and the outcome being sought.
“Root cause” can be an intellectually misleading metaphor
The language of root causes suggests that a complex phenomenon resembles a plant: visible outcomes appear above ground, while one deeper cause sits underneath and generates everything else. In some scientific problems, identifying a major upstream cause is genuinely possible and useful. Human behaviour often does not conform so neatly to this structure.
Causes can form networks, feedback loops and developmental sequences. An environmental condition can alter behaviour, which changes the environment, which then reinforces the behaviour. Psychological expectations can shape attention, and selective attention can provide evidence that strengthens the expectation. Social reactions can influence physiology, while physiological state changes how social information is perceived.
In such cases, asking for the root cause may impose a causal architecture before the evidence has established one. The more informative questions concern which factors are necessary, sufficient, contributing, maintaining or moderating, and how their influence changes over time.
This language is less dramatic than identifying a hidden root cause, but it is scientifically more precise.
Research literacy requires knowing what a study can actually explain
Different research designs support different explanatory claims. Cross-sectional studies can identify associations but generally provide limited evidence about temporal order and causality. Longitudinal research can establish sequence more effectively but may still leave confounding variables unresolved. Experimental manipulation can strengthen causal inference, although experiments often simplify conditions and may not generalise straightforwardly to complex real-world environments.
Measurement also determines what can be concluded. A study that measures self-reported anxiety and avoidance cannot directly establish a neural mechanism. A brain-imaging study can identify patterns associated with a task without automatically telling us how those patterns relate to everyday behaviour. A qualitative study can provide rich information about meaning and experience without estimating population-level causal effects.
These are not deficiencies that make one methodology superior to another. Different methods are designed to answer different questions, and scientific understanding often improves when evidence from methods with different strengths converges.
Problems arise when conclusions migrate beyond the design that produced them. An association becomes described as a cause, a laboratory effect becomes a universal mechanism, or a biological correlate becomes a comprehensive explanation of a psychological phenomenon. Research literacy involves noticing those transitions.
More levels do not always mean better science
An article arguing against single-level explanation must also resist the opposite excess. Not every research question benefits from including psychological, physiological, relational, behavioural, developmental, cultural and environmental variables simultaneously.
Adding levels increases complexity. It creates additional measurement problems, more possible interactions and greater opportunities for post hoc interpretation. A narrow, well-defined question answered rigorously can be more scientifically valuable than an ambitious “whole-person” model supported only weakly.
The relevant standard is therefore explanatory need. If a behavioural account explains the phenomenon under investigation and the question does not require neural detail, adding neuroscience may contribute little. If a social intervention reliably changes an outcome, molecular measurements are not automatically necessary to establish that the intervention works. Conversely, if competing psychological theories make identical behavioural predictions but different neural predictions, biological measurement may become highly informative.
Pluralism should therefore be selective rather than ceremonial. Multiple levels are useful when they solve explanatory problems, constrain competing accounts or reveal relationships that would otherwise remain invisible.
The aim is better discrimination, not maximum complexity
The strongest reason for using multiple explanations is not that humans are mysterious or infinitely complicated. It is that better distinctions produce better questions.
A behavioural description may identify what is repeated. A psychological account may clarify what is anticipated or represented. A relational analysis may show why the same person behaves differently with different people. Environmental analysis may identify the conditions that make particular actions easy, difficult, rewarding or costly. Physiological investigation may reveal mechanisms and constraints through which those processes are implemented.
Each level can add something without automatically cancelling what came before it. At the same time, each explanation remains accountable to evidence and should be limited to the question it can genuinely answer.
This way of thinking also changes how disagreement is approached. Two researchers may appear to disagree about why a behaviour occurs when they are actually answering different questions. Conversely, two accounts may appear complementary because they operate at different levels while making genuinely incompatible causal claims. Knowing the difference is part of scientific literacy.
The question should therefore rarely be, Which explanation is the real one? A more productive set of questions asks what phenomenon is being explained, which aspect of it the explanation addresses, what evidence supports the proposed relationship, what remains unexplained, and whether another level contributes information that cannot be obtained from the first.
One explanation is rarely enough because human behaviour exists simultaneously as biological activity, psychological functioning, observable behaviour, relational interaction and action within physical and social environments. Those descriptions do not need to compete for ownership of the person. The scientific task is to determine when each one matters, how they constrain one another, and how much explanatory weight the available evidence actually allows each of them to carry.
Leave a Reply