Imagine putting off a difficult task and feeling your shoulders relax. For the moment, the decision seems to have worked. Tomorrow, however, the unfinished task may require more effort, and someone else may be waiting for your contribution. The relief was real. So are the consequences that arrive later. Which part of this experience should guide the next decision?

Consequences bring our intentions into contact with what our actions actually produce. They can reveal that an approach is effective, that a familiar habit carries costs, or that a situation differs from what we expected. Yet consequences also need interpretation. We may repeat what provides immediate comfort, misidentify the cause of a success, or abandon a reasonable approach after one disappointing result (Baron & Hershey, 1988; Skinner, 1953).

Learning from correlations means attending to recurring relationships between events. The harder task is judging what those relationships tell us, how circumstances shape them, and where our explanation remains uncertain. Behavioral learning, systems theory, and research on judgment offer complementary ways to understand this task. Together, they support a practical form of agency: using consequences to improve our responses while recognizing that we never control every influence on an outcome (Bandura, 1999; Meadows, 2008).

What It Means to Learn from Consequences

A consequence is an effect or outcome of an action. In behavioral learning, consequences matter because they can change what an individual does next. We can also deliberately examine outcomes to revise our expectations and choices. These processes overlap, but they are distinguishable: behavior may change without a conscious explanation, while a person may understand a cost and still struggle to change the behavior that produces it (Bandura, 1999; Skinner, 1953).

An observed event is not automatically a consequence of the action that preceded it. A conversation followed by improved cooperation may have helped, but a change in workload or another person’s circumstances may also explain the improvement. Correlation describes how events or conditions vary together; causation concerns whether one factor helps produce a change in another. The distinction protects the value of experience by keeping us from asking it to establish more than it can (Cheng, 1997).

Systems theory cautions against assuming that every relationship is linear or that the same action will have the same effect under changing conditions (Sterman, 1994; von Bertalanffy, 1968). Causal learning adds a related caution: an observed association needs to be evaluated against alternative causes before it can support a causal explanation (Cheng, 1997). Patterns can guide investigation without becoming guarantees.

How Consequences Shape Behavior

Reinforcement and the Persistence of Behavior

Skinner placed the consequences of action at the center of operant learning. In operant conditioning, behavior acts on the environment, and its consequences alter the likelihood of similar behavior occurring again. A response that produces access to something valued, or escape from something aversive, may become more frequent. The relevant question is what the consequence actually does to behavior under the circumstances (Skinner, 1953).

Behavior reinforcement therefore has a precise meaning. Positive reinforcement involves adding a stimulus that increases behavior; negative reinforcement involves removing or reducing an aversive stimulus in a way that increases behavior. The words positive and negative refer to addition and removal, not to moral value. An intended reward may fail to reinforce. Skinner distinguished these processes from punishment and cautioned that suppressing a response immediately does not necessarily produce a lasting reduction in that behavior (Skinner, 1953, pp. 72–73, 182–186).

This helps explain the opening example. If postponing a task repeatedly brings relief and makes postponement more likely, relief is functioning as negative reinforcement. The later cost does not make the immediate effect imaginary. It shows that the behavior has consequences operating on different time scales. Understanding the habit requires noticing both, rather than assuming that awareness of future trouble should be sufficient to stop it.

Causal Learning and Background Conditions

To understand what an event tells us, we need to notice what happens in its absence. Rescorla studied this by comparing how rats responded to a tone when shocks occurred both during and outside it. When shocks were equally likely with or without the tone, the tone did not produce the fear response measured in the experiment—even though it sometimes accompanied a shock. The tone added no useful information about when a shock would occur (Rescorla, 1968).

The study concerns Pavlovian learning, where a cue predicts an event, rather than operant learning, where an action produces a consequence. Its relevance is the importance of comparison. Seeing an outcome after an event is less informative when that outcome happens just as often without it. Everyday causal reasoning requires a further step: considering other influences and the conditions under which an apparent relationship would support a causal explanation (Cheng, 1997).

Cheng’s causal power theory develops this distinction by separating the frequency of an outcome from a factor’s capacity to produce or prevent it. An outcome may be common because other causes are already producing it. If those causes occur more often when the action of interest occurs, the apparent benefit of that action may be misleading. For example, a new planning routine introduced during a quieter work period may appear effective partly because demands have fallen. Meaningful comparisons need to address that changing background, alongside the action itself (Cheng, 1997).

A comparison can also tell us too little when success is already certain. Imagine a team that completes every task on time before trying a new planning routine. Its completion rate cannot rise further, so that measure alone cannot show whether the routine helps. Cheng describes this as a ceiling that can hide a causal contribution. An unchanged result does not prove that the routine is useless, but it does not establish a benefit either (Cheng, 1997).

A Systems View of Consequences

An action rarely enters an otherwise empty world. It enters a household, workplace, or community with an existing history, distribution of resources, and pattern of relationships. General systems theory directs attention to these connections. Von Bertalanffy emphasized that the behavior of an organized whole cannot always be understood by studying its parts in isolation. Meadows similarly locates recurring behavior in the structure and interconnections of a system (Meadows, 2008; von Bertalanffy, 1968).

Feedback Loops and Delayed Effects

A feedback loop occurs when the effects of a process return to influence its subsequent operation. A reinforcing loop amplifies change; a balancing loop counteracts change or moves a system toward a goal. Neither term tells us whether the result is desirable. A balancing process may maintain an unhealthy pattern, while a reinforcing process may support growing competence. These system-level meanings differ from the behavioral distinction between positive and negative reinforcement (Meadows, 2008).

Consider a hypothetical team that falls behind and responds by repeatedly postponing planning. More time becomes available for immediate tasks, but coordination may deteriorate. Rework then adds to the backlog, creating further pressure to skip planning. The decision is understandable at each moment, yet the pattern may sustain the problem. This is a systems interpretation of the example, illustrating how a local benefit can be accompanied by costs elsewhere in the process (Meadows, 2008).

Delays make these patterns harder to read. Today’s improvement may reflect work completed earlier, while today’s decision may not show its full effects for weeks. Meadows describes stocks as accumulations that change through flows over time. A backlog is a straightforward example: it depends on the relationship between incoming and completed work, not simply on how busy everyone feels today. Evaluating consequences requires a time window suited to the process being examined (Meadows, 2008).

Sterman explains how delays can also change the consequences of our attempts to correct a problem. When the effects of an earlier response remain unseen, we may intervene again as though nothing has been done. The accumulated corrections can then overshoot the goal. Delays also reduce opportunities to compare expectations with results before circumstances change. In the team example, recognizing work already underway would help distinguish an insufficient response from a response whose effects have not yet arrived (Sterman, 1994).

Diehl and Sterman demonstrated the difficulty of delayed feedback in an inventory-management experiment. Participants had incentives and opportunities to learn, yet their performance deteriorated as time delays and feedback effects increased. The setting was a controlled management task, so it does not establish how every personal decision unfolds. It does show why observing results can be insufficient when the connection between action and outcome is difficult to track (Diehl & Sterman, 1995).

Learning in Complex and Changing Systems

In complex systems, the parts affect one another, and the overall pattern grows out of those connections. Page explains why we cannot understand that pattern simply by examining each part on its own. In systems that adapt, people or other participants can also change their responses. Axelrod and Cohen describe how different approaches arise, interact, and become more or less common as some are favored over others. As these patterns change, an approach that worked before may no longer produce the same result (Axelrod & Cohen, 1999; Page, 2011).

Nonlinear relationships are another source of difficulty. A change in one factor need not produce a proportionate change in another. A team may absorb a few additional requests while it has spare capacity, then accumulate a backlog once incoming work exceeds what it can complete. Understanding this shift requires attention to how much work the team can complete and how unfinished work accumulates. Systems concepts become useful when they identify such relationships clearly enough to examine (Meadows, 2008; von Bertalanffy, 1968).

Our own participation also changes the setting. Bandura’s reciprocal determinism describes personal factors, behavior, and environment as interacting influences. Asking a clearer question may change the response we receive; that response may change our expectations and our next question. Influence need not be equal or simultaneous. Other people retain their own purposes, and broader conditions constrain what any participant can accomplish (Bandura, 1999).

History matters as well. Bouton’s review of Pavlovian extinction research shows that reduced conditioned responding often involves new learning whose expression depends on context, rather than the complete loss of earlier learning (Bouton, 2004). A changed response in one setting does not necessarily demonstrate that previous learning has disappeared. Applied cautiously to everyday learning, this finding invites us to examine whether a change persists across settings.

Unexpected outcomes consequently deserve investigation. Some reflect chance variation; others expose a missing condition, a changed environment, or a mistaken explanation. Complexity is not a reason to protect a favored belief from contrary evidence. It is a reason to examine more carefully what the belief predicts and where it stops being useful (Page, 2011).

Outcome Bias and the Limits of Experience

A favorable result can make an earlier decision appear better than it was. Across five studies, Baron and Hershey found that people evaluated decisions more favorably when outcomes were favorable, even when they were given the relevant information available to the decision-maker. This outcome bias makes it difficult to separate the quality of the reasoning from the good or bad fortune that followed (Baron & Hershey, 1988).

For our own lives, this distinction permits a more useful review. We can acknowledge the outcome and still ask whether the decision took reasonable account of what was known at the time. A single success should not establish a strategy’s value, and a single failure need not establish its worthlessness. Repeated patterns remain relevant, especially when they contradict our expectations. The aim is to learn from results without letting one result rewrite the entire decision.

The quality of the learning environment also matters. Kahneman and Klein identify two conditions for acquiring skilled intuition: sufficiently predictable relationships in the environment and opportunities to learn those relationships. They also caution that subjective confidence is an unreliable indicator of accuracy (Kahneman & Klein, 2009). Experience in a setting with delayed, ambiguous, or inconsistent feedback may therefore produce confidence without comparable skill. Familiarity with a problem deserves consideration alongside the conditions under which that familiarity developed.

Self-Observation and the Accuracy of Confidence

Looking inward can reveal that a task feels threatening or that an interaction leaves us relieved. Explaining why we acted, or why the other person responded, requires additional interpretation. We should not discard the felt experience, but we can distinguish it from the causal account we construct around it. That distinction allows self-awareness to contribute evidence while remaining open to correction (Schwitzgebel, 2004).

Schwitzgebel’s examination of Titchener’s laboratory manual treats introspective training as a possibility worth exploring while identifying serious difficulties in establishing whether it improves accuracy. Agreement among trained observers could reflect improved observation, but training could also shape expectations and reports. The difficulty is finding a way to tell improved accuracy from shared assumptions, rather than treating agreement itself as proof of insight (Schwitzgebel, 2004).

Fleming and colleagues studied a specific kind of self-knowledge: how well people could tell when their own answers were right. In a study of 32 healthy participants making visual judgments, some people’s confidence tracked the accuracy of their answers more closely than others’, even though their overall task performance was similar. This ability, called metacognitive sensitivity, was associated with differences in brain structure, including areas toward the front of the brain and their connections. The study did not show that those brain differences caused the ability. Nor did it test insight into relationships, motives, or life choices (Fleming et al., 2010).

Taken together, the introspection papers suggest a modest principle for reflection: confidence about our interpretation deserves comparison with evidence (Fleming et al., 2010; Schwitzgebel, 2004). As a practical application, we can notice what we felt, specify what we expected, and later examine what occurred. Another person’s account may reveal something we missed, although their interpretation also deserves scrutiny. This is a way to approach reflection, rather than a self-reflection method tested by these studies.

Learning More Carefully from Consequences

Defining Meaningful Outcomes

Before judging whether an action worked, we need some sense of what success means. Axelrod and Cohen emphasize that the criteria used to select successful agents or strategies shape how a system adapts. If a workplace measures only immediate output, activities that protect future quality may look unproductive. If we evaluate a difficult conversation only by whether tension disappeared, avoidance may look successful even when the issue remains (Axelrod & Cohen, 1999).

This invites a wider view of consequences. Immediate relief may matter alongside completed work, the effect on another person, and the sustainability of the response. These considerations need not point in the same direction. Defining the outcome more carefully makes the disagreement visible: we may be choosing between competing aims, rather than failing to discover one universally correct behavior (Axelrod & Cohen, 1999; Meadows, 2008).

Comparing Outcomes and Revising Mental Models

A useful record separates an expectation from its eventual result. For the postponed task, this might mean noting the anticipated difficulty, what was actually attempted, and what happened over the following days. Comparisons become more informative when we also notice workload, available support, and other changes. Such observations can suggest explanations worth testing. Informal tracking alone cannot rule out other influences or establish causation (Cheng, 1997).

In his discussion of Chris Argyris’s distinction between single-loop and double-loop learning, Sterman explains how we can change a decision while leaving our goals and explanation of the problem intact. Deeper revision changes the mental model itself: which influences we include, how far ahead we look, and what we regard as success (Sterman, 1994, pp. 294–297). A team might respond to a backlog by demanding faster work, or reconsider whether poor coordination and excessive incoming demands are sustaining it. These different explanations lead to different responses.

One practical application of these ideas is to try a small change that can be reversed if needed. When circumstances allow, this can make it easier to observe what follows and consider possible explanations (Axelrod & Cohen, 1999; Cheng, 1997). Trying a shorter work session, clarifying one responsibility, or changing when a discussion occurs may help reveal which conditions matter. The point is to make an adjustment understandable enough to learn from. If several things change together, the improvement may be welcome even while its cause remains uncertain.

Variation also preserves opportunities to adapt. Axelrod and Cohen discuss the tension between using approaches that currently perform well and exploring alternatives. Page shows why diversity can contribute to a system’s ability to maintain its functioning, while cautioning that more diversity is not invariably beneficial. Applied to personal learning, this supports keeping workable alternatives available without changing everything after each disappointment (Axelrod & Cohen, 1999; Page, 2011).

Agency Without Complete Control

Learning from consequences can strengthen agency without making the individual responsible for every outcome. Bandura describes agency as operating within social structures and through personal, proxy, and collective action. Sometimes a better response involves a skill we can practice. Sometimes it involves asking for assistance or working with others to alter conditions that one person cannot change alone (Bandura, 1999).

An unfavorable outcome does not, by itself, establish a failure of effort or character. Bandura emphasizes that social structures can both support and constrain action, while research on outcome bias cautions against judging a decision solely by its result (Bandura, 1999; Baron & Hershey, 1988). For questions of responsibility, this suggests examining the choices and resources actually available. Recognizing constraints need not erase accountability for harm; it can make that assessment more careful.

Consequences provide evidence about how our lives are unfolding. They cannot, on their own, settle what we should value or how we should treat one another. A behavior that reliably produces an advantage can still impose an unacceptable cost on someone else. Reflective learning therefore asks both whether an approach achieves its aim and whether that aim and its costs deserve our continued support.

A Few Words by Psychology Fanatic

The unfinished task may still be waiting tomorrow. We can meet it with curiosity about the relief that made postponement appealing and honesty about the burden that followed. Perhaps the next useful step is beginning. Perhaps it is asking for help, reducing an unrealistic demand, or acknowledging that the task itself needs to change.

We do not need a flawless explanation before making a thoughtful adjustment. We do need enough openness to notice what the adjustment brings. The value of consequences lies partly in their capacity to surprise us, interrupt a familiar story, and invite a more accurate response. We can take that invitation seriously without treating every disappointment as a verdict on who we are.

Associated Concepts

• Behavior Reinforcement: Reinforcement occurs when a consequence increases the likelihood of a behavior recurring. It helps explain why immediate relief or reward can sustain a pattern even when its later effects are costly.
• Feedback Loops: Feedback loops occur when the effects of a process return to influence how it continues. They help explain how our responses can amplify change, maintain familiar patterns, or produce consequences that emerge over time.
• Reciprocal Determinism: Albert Bandura described personal factors, behavior, and environment as interacting influences. This perspective places learning within an ongoing exchange: our actions help shape the circumstances that, in turn, influence what we do next.
• Self-Awareness: Self-awareness involves noticing our thoughts, feelings, and behavior. It contributes to learning from consequences when we distinguish what we experience from the explanations we form, then compare those explanations with evidence.
• Complex Systems: Complex systems contain interacting parts whose connections produce patterns that cannot be understood by examining each part alone. This perspective helps explain why an action’s consequences depend on surrounding conditions and may change as the system adapts.
• Metacognition: Metacognition involves monitoring and regulating our own thinking. It supports reflective learning by directing attention to how we reach conclusions, how confident we feel, and whether that confidence is justified by the available evidence.

References

Axelrod, Robert; Cohen, Michael D. (1999). Harnessing Complexity: Organizational Implications of a Scientific Frontier. Free Press. ISBN: 9780684867175.
(Return to Main Text)

Bandura, Albert (1999). Social Cognitive Theory: An Agentic Perspective. Asian Journal of Social Psychology, 2(1), 21–41. DOI: 10.1111/1467-839X.00024.
(Return to Main Text)

Baron, Jonathan; Hershey, John C. (1988). Outcome Bias in Decision Evaluation. Journal of Personality and Social Psychology, 54(4), 569–579. DOI: 10.1037/0022-3514.54.4.569.
(Return to Main Text)

Bouton, Mark E. (2004). Context and Behavioral Processes in Extinction. Learning & Memory, 11(5), 485–494. DOI: 10.1101/lm.78804.
(Return to Main Text)

Cheng, Patricia W. (1997). From Covariation to Causation: A Causal Power Theory. Psychological Review, 104(2), 367–405. DOI: 10.1037/0033-295X.104.2.367.
(Return to Main Text)

Diehl, Ernst; Sterman, John D. (1995). Effects of Feedback Complexity on Dynamic Decision Making. Organizational Behavior and Human Decision Processes, 62(2), 198–215. DOI: 10.1006/obhd.1995.1043.
(Return to Main Text)

Fleming, Stephen M.; Weil, Rimona S.; Nagy, Zoltan; Dolan, Raymond J.; Rees, Geraint (2010). Relating Introspective Accuracy to Individual Differences in Brain Structure. Science, 329(5998), 1541–1543. DOI: 10.1126/science.1191883.
(Return to Main Text)

Kahneman, Daniel; Klein, Gary (2009). Conditions for Intuitive Expertise: A Failure to Disagree. American Psychologist, 64(6), 515–526. DOI: 10.1037/a0016755.
(Return to Main Text)

Meadows, Donella H. (2008). Thinking in Systems: A Primer. Edited by Diana Wright. Chelsea Green Publishing. ISBN: 9781603580557.
(Return to Main Text)

Page, Scott E. (2011). Diversity and Complexity. Princeton University Press. ISBN: 9780691137674.
(Return to Main Text)

Rescorla, Robert A. (1968). Probability of Shock in the Presence and Absence of CS in Fear Conditioning. Journal of Comparative and Physiological Psychology, 66(1), 1–5. DOI: 10.1037/h0025984.
(Return to Main Text)

Schwitzgebel, Eric (2004). Introspective Training Apprehensively Defended: Reflections on Titchener’s Lab Manual. Journal of Consciousness Studies, 11(7–8), 58–76. Website: Publication record.
(Return to Main Text)

Skinner, Burrhus Frederic (1953). Science and Human Behavior. Macmillan. Website: Publication record.
(Return to Main Text)

Sterman, John D. (1994). Learning in and about Complex Systems. System Dynamics Review, 10(2–3), 291–330. DOI: 10.1002/sdr.4260100214.
(Return to Main Text)

von Bertalanffy, Ludwig (1968). General System Theory: Foundations, Development, Applications. George Braziller.
(Return to Main Text)