Entrepreneurial Action and Learning Under Uncertainty

INDEPENDENT SCHOLARSHIP · Not peer-reviewed

Cody Thomas Rounds, M.A.

Clinical Psychologist-Master

The decision problem behind an entrepreneurial mindset

Entrepreneurs often have to act without knowing whether a market will respond, a partner will remain available, or a planned service can be delivered at acceptable cost. Advice to become more confident does not answer the decision problem. Confidence can support action, but it can also protect an untested explanation from correction. The useful question is how to make a commitment that is both consequential enough to teach something and bounded enough to preserve future options.

I argue that the entrepreneurial mindset should be analyzed through the relationship between action and learning. The central capacity is to identify what remains uncertain, choose an informative action, interpret its consequences, and revise the next commitment. This account allows persistence, caution, and termination to be adaptive under different conditions. None is inherently evidence of entrepreneurial strength or weakness.

The argument concerns decision processes rather than a universal entrepreneurial personality. It also distinguishes business performance from psychological health. A sound decision can be followed by an unfavorable outcome, and a profitable venture can impose serious personal costs. Neither financial success nor comfort during uncertainty is a sufficient measure of the quality of the process that produced it.

Uncertainty and the limits of calculation

Knight distinguished measurable risk from uncertainty that cannot be represented adequately through known probabilities (Knight, 1921). The distinction remains useful because an early venture may lack the stable categories, frequencies, or market conditions needed for precise forecasting. This does not make analysis pointless. It changes what analysis can reasonably claim and increases the importance of actions that reveal information or create new possibilities.

Bounded rationality further recognizes that decision makers operate with limits in information, time, and computational capacity (Simon, 1955). A founder cannot evaluate every possible product, customer, and organizational arrangement before acting. Criteria for an acceptable next step may therefore be more useful than an attempt to identify an unknowable optimum. The standard is a reasoned choice under constraints, not the elimination of uncertainty.

Entrepreneurship theory has examined how uncertainty becomes relevant to an actor's willingness and ability to act (McMullen & Shepherd, 2006). Opportunity beliefs also develop through interpretation and updating, rather than arrive as fully formed discoveries (Shepherd et al., 2007). These perspectives make the founder's judgment central without treating belief as evidence of market demand. A proposition can be personally compelling and commercially unsupported at the same time.

Planning effectuation and organizational learning

Effectuation offers an account of beginning with available means, limiting affordable loss, and working with emerging commitments rather than relying entirely on prediction of a predetermined outcome (Sarasvathy, 2001). It should not be reduced to improvising without discipline. Nor should it be treated as a general psychological virtue that makes planning obsolete. Some venture questions permit useful prediction; others require action before the relevant information exists.

A meta analytic review associated dimensions of effectuation with venture performance (Read et al., 2009). Such associations do not establish that adopting one decision rule will cause improvement in every venture. A qualitative comparison of six emerging ventures identified overlaps and differences among effectuation, causation, and bricolage (Fisher, 2012). The findings support attention to what people do in context rather than forcing each founder into one permanent style.

Organizational learning adds a tension between exploring possibilities and exploiting what is already known. Exploration can generate new options but consume resources without immediate return; exploitation can improve efficiency while narrowing future adaptation (March, 1991). The practical challenge is to allocate attention and commitment across these activities. Endless experimentation can become avoidance of delivery, while relentless execution can entrench an incorrect premise.

An action and feedback account therefore needs both openness and discipline. The founder must remain willing to revise assumptions, but also keep an action stable long enough to produce interpretable information. Changing the offer, audience, price, and delivery method after every response can make it impossible to know what was learned. The aim is purposeful variation within a decision process, not movement for its own sake.

What intervention studies actually show

An initial randomized trial with 116 Italian startups examined training in a more scientific approach to developing predictions and testing ideas. Both groups received general business training, while the intervention group received additional guidance in explicit hypotheses and rigorous tests. The study reported favorable differences in performance and strategic change (Camuffo et al., 2020). It supports the teachability of particular practices in that setting, not a universal claim that every entrepreneur should replicate a laboratory scientist's methods.

A larger replication and extension, involving 759 firms across four randomized trials, sharpened the picture. The scientific approach increased idea termination and produced a more nuanced pattern of radical pivots, favoring some strategic change over either no change or repeated change (Camuffo et al., 2024). One important implication is that ending an idea can be a productive result of training. A program that helps entrepreneurs stop weak projects should not be judged solely by whether more initial ideas survive.

Other research broadens the evidence without testing the same intervention. A randomized trial with 1,500 microenterprise owners in Togo found that personal initiative training improved profits relative to control, while the estimated gain from traditional business training was smaller and not statistically significant (Campos et al., 2017). The program combined psychological ideas with entrepreneurial behavior. Its results do not show that a positive mindset alone improves every business or that findings transport automatically across economic settings.

Research on adoption of A/B testing technology among high technology startups also linked experimentation with performance and learning (Koning et al., 2022). That study used observational adoption data rather than random assignment to a general entrepreneurial mindset. Digital environments can generate large volumes of rapid feedback at relatively low cost. A professional service or capital intensive venture may face slower, smaller, and more ambiguous tests. The decision principle can be relevant while the available methods differ substantially.

An action and feedback framework

The following framework synthesizes ideas from these literatures. Its contribution is to connect the uncertainty being addressed with the commitment, observation, and subsequent decision. It is not an independently validated training method. The steps can be brief when a decision is reversible and more explicit when consequences are substantial.

Specify the uncertainty that matters next

The first task is to identify which assumption could change the next commitment. A founder may need to know whether a particular customer has the problem, whether the customer will pay for the proposed solution, or whether the service can be delivered reliably. These are different questions. General enthusiasm for an idea answers none of them conclusively.

The uncertainty should be stated at a useful level. “Will the business succeed?” is too broad for one test. “Will members of this defined group commit to a paid pilot under these terms?” gives an observable question, although its result will still have limits. The founder should distinguish an uncertain fact from a decision about values, such as how much personal time a venture is worth.

Choose a bounded and informative commitment

A useful action exposes the assumption to relevant evidence while keeping losses within acceptable limits. A pilot, a prototype, a narrowly defined service, or a structured purchasing conversation may be appropriate depending on the question. Affordable loss includes more than cash. Time, reputation, obligations to others, and the ability to recover from a failed attempt also matter.

The commitment must be sufficient to generate interpretable evidence. An offer never presented to an appropriate audience cannot test demand. A prototype that omits the feature people are expected to value may test a different proposition. Equally, a large investment should not be described as an experiment merely because its outcome is uncertain. The action needs an explicit learning purpose and a credible route to revision.

State expectations before interpreting results

Before acting, the founder should record what observations would increase or decrease confidence in the assumption. Perfect precision is rarely available, but a prior statement makes later rationalization easier to detect. It also clarifies what the test cannot resolve. A small pilot may reveal delivery problems while remaining uninformative about a broad market.

Alternative explanations should be considered in advance. A lack of purchase could reflect price, audience mismatch, weak communication, timing, or lack of need. A purchase could reflect friendship or a one time curiosity rather than durable demand. The plan need not eliminate every ambiguity. It should identify which interpretations the available observation can reasonably distinguish.

Obtain feedback from the relevant behavior

The most accessible feedback is not always the most useful. Compliments, survey interest, website visits, purchases, repeat use, and successful delivery refer to different points in a proposed business model. Each can contribute, but their meaning depends on the decision. Treating all favorable responses as validation allows a founder to accumulate encouragement without examining the uncertain link.

Feedback also has a social context. Team members may withhold concerns if disagreement is treated as disloyalty. Research on psychological safety associated a climate supportive of interpersonal risk taking with team learning behavior (Edmondson, 1999). This is related organizational evidence, not direct validation of the framework. Its implication is that a founder's response to unwelcome information can affect whether useful information is offered at all.

Interpret the result and decide what changes

Interpretation should return to the original question, the quality of the test, and plausible alternatives. A favorable result may justify a larger commitment rather than a claim that the whole business model is established. An unfavorable result may warrant changing the offer, conducting a different test, or ending the idea. An ambiguous result may require acknowledging that little was learned.

The next decision should be specific. “Keep going” leaves the scope and exposure unclear. Continuing the same bounded offer for another defined period differs from signing a long lease. The founder should explain why the evidence supports the size of the next commitment. Learning becomes practically useful through that change in action.

Review stopping and personal consequences

Stopping rules deserve consideration before resources and identity become deeply invested. They can specify a maximum exposure, a failed assumption, an unacceptable burden, or a condition requiring independent review. Rules need not be rigid when circumstances change, but revising them should require an explicit reason. Otherwise every threshold can move just before it is crossed.

Personal consequences should be evaluated separately from commercial signals. Research on goal adjustment associates disengagement from unattainable goals and reengagement elsewhere with well being (Wrosch et al., 2003). This does not prescribe abandonment whenever work becomes difficult. It supports examining whether a valued direction can continue through a different commitment, and whether preserving one idea has become more important than the broader purpose it was meant to serve.

A hypothetical service venture

Consider a constructed example of a founder planning a professional development service. Early conversations produce enthusiastic responses, and the founder interprets them as evidence of demand. The proposed next step is a large investment in a complete program. The action and feedback framework begins by asking what the conversations actually established.

The contacts may recognize the problem and like the idea. They have not necessarily demonstrated purchasing authority, willingness to pay, or readiness to allocate employee time. The founder identifies a more consequential next uncertainty: whether a defined buyer group will commit to a clearly scoped paid pilot. The offer states its purpose and limits accurately, and the founder ensures that the promised service can be delivered.

Before presenting the offer, the founder records what level and kind of response would justify another step, how much time and money can be committed, and what reasons for refusal would matter. A friendly contact's encouragement is distinguished from an authorized purchase. The founder also records which aspects remain untested, including repeat demand and the feasibility of delivering at larger scale.

Suppose several buyers express interest but none purchases. That result does not establish that the whole market lacks need. It does weaken the inference that interest under the initial conditions is sufficient for a sale. Further discussion reveals that the people approached lack budget authority and that the timing conflicts with their planning cycle. A revised test can address the audience and timing, provided that the founder does not use each disappointing result to preserve the original belief indefinitely.

Now suppose a pilot is purchased and participants report satisfaction, but delivery requires far more preparation than anticipated. The demand assumption has received some support; the delivery and cost assumptions have not. Scaling immediately could amplify the problem. The next decision may concern a narrower service, a different delivery model, a revised price, or termination of this version of the offer.

The example shows why one global label such as “validated idea” is inadequate. The founder learns about distinct links in sequence and keeps the next commitment proportional to that learning. An apparently disappointing result can improve the decision, while a favorable response can leave the most consequential uncertainty unresolved.

Bias identity and escalation of commitment

The ability to revise a decision becomes harder when the venture is bound to a founder's identity or public reputation. Classic research demonstrated escalation following personal responsibility for an initial investment (Staw, 1976). Research on sunk costs likewise showed that prior expenditures can influence later choices even when those expenditures cannot be recovered (Arkes & Blumer, 1985). A meta analytic review identified multiple determinants of escalation rather than one simple personality explanation (Sleesman et al., 2012).

The practical implication is to design review before commitment becomes emotionally expensive. An independent discussion, an explicit account of alternatives, or a decision record can make it easier to ask whether the next investment is justified on its own terms. These practices do not eliminate bias. Their purpose is to create occasions when evidence can interrupt the momentum of the original choice.

Experience also requires appraisal. Expert intuition is more dependable when environments contain learnable regularities and the person has opportunities for informative feedback (Kahneman & Klein, 2009). Confidence developed in one market may travel poorly to another. A founder's strong feeling can be useful information about an impression without being reliable evidence that the impression is correct.

Limits and implications

The framework is least straightforward where feedback is delayed, outcomes are rare, or a test changes the environment being examined. Small samples may provide little statistical precision, and ethical or contractual obligations can limit experimentation. Some decisions are partly irreversible. These conditions require more careful judgment and consultation, not an insistence that every problem can be solved through a cheap experiment.

The evidence reviewed also spans different settings and levels of analysis. Studies of startup training, microenterprise profits, digital testing, and organizational teams cannot be collapsed into one effect estimate. The proposed integration needs direct evaluation, including whether decision records improve the quality of revision, whether termination becomes more appropriately timed, and whether burdens or harms increase. Venture survival alone would be an inadequate outcome because ending a weak idea may be part of better learning.

Conclusion

Entrepreneurial action can be understood as a sequence of commitments that changes what a person can know and do next. The quality of that sequence depends on explicit assumptions, informative feedback, and a willingness to revise the scale or direction of investment. This account gives psychological capacities a specific role without reducing entrepreneurship to confidence or generic resilience. An adaptive mindset makes commitment responsive to learning while keeping commercial performance and personal well being available for separate review.

References

Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124–140. doi:10.1016/0749-5978(85)90049-4

Campos, F., Frese, M., Goldstein, M., Iacovone, L., Johnson, H. C., McKenzie, D., & Mensmann, M. (2017). Teaching personal initiative beats traditional training in boosting small business in West Africa. Science, 357(6357), 1287–1290. doi:10.1126/science.aan5329

Camuffo, A., Cordova, A., Gambardella, A., & Spina, C. (2020). A scientific approach to entrepreneurial decision making: Evidence from a randomized control trial. Management Science, 66(2), 564–586. doi:10.1287/mnsc.2018.3249

Camuffo, A., Gambardella, A., Messinese, D., Novelli, E., Paolucci, E., & Spina, C. (2024). A scientific approach to entrepreneurial decision‐making: Large‐scale replication and extension. Strategic Management Journal, 45(6), 1209–1237. doi:10.1002/smj.3580

Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. doi:10.2307/2666999

Fisher, G. (2012). Effectuation, causation, and bricolage: A behavioral comparison of emerging theories in entrepreneurship research. Entrepreneurship Theory and Practice, 36(5), 1019–1051. doi:10.1111/j.1540-6520.2012.00537.x

Kahneman, D., & Klein, G. (2009). Conditions for intuitive expertise: A failure to disagree. American Psychologist, 64(6), 515–526. doi:10.1037/a0016755

Knight, F. H. (1921). Risk, uncertainty and profit. Houghton Mifflin.

Koning, R., Hasan, S., & Chatterji, A. (2022). Experimentation and start-up performance: Evidence from A/B testing. Management Science, 68(9), 6434–6453. doi:10.1287/mnsc.2021.4209

March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2(1), 71–87. doi:10.1287/orsc.2.1.71

McMullen, J. S., & Shepherd, D. A. (2006). Entrepreneurial action and the role of uncertainty in the theory of the entrepreneur. Academy of Management Review, 31(1), 132–152. doi:10.5465/amr.2006.19379628

Read, S., Song, M., & Smit, W. (2009). A meta-analytic review of effectuation and venture performance. Journal of Business Venturing, 24(6), 573–587. doi:10.1016/j.jbusvent.2008.02.005

Sarasvathy, S. D. (2001). Causation and effectuation: Toward a theoretical shift from economic inevitability to entrepreneurial contingency. The Academy of Management Review, 26(2), 243–263. doi:10.2307/259121

Shepherd, D. A., McMullen, J. S., & Jennings, P. D. (2007). The formation of opportunity beliefs: Overcoming ignorance and reducing doubt. Strategic Entrepreneurship Journal, 1(1-2), 75–95. doi:10.1002/sej.3

Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics, 69(1), 99–118. doi:10.2307/1884852

Sleesman, D. J., Conlon, D. E., McNamara, G., & Miles, J. E. (2012). Cleaning up the big muddy: A meta-analytic review of the determinants of escalation of commitment. Academy of Management Journal, 55(3), 541–562. doi:10.5465/amj.2010.0696

Staw, B. M. (1976). Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance, 16(1), 27–44. doi:10.1016/0030-5073(76)90005-2

Wrosch, C., Scheier, M. F., Miller, G. E., Schulz, R., & Carver, C. S. (2003). Adaptive self-regulation of unattainable goals: Goal disengagement, goal reengagement, and subjective well-being. Personality and Social Psychology Bulletin, 29(12), 1494–1508. doi:10.1177/0146167203256921

AI Use Disclosure

Artificial intelligence tools were used throughout the development and preparation of this manuscript to support research, critical analysis, writing, revision, and editorial review. ChatGPT-4 Clinicians (OpenAI) was used to support clinical research exploration, literature review, critical reflection, conceptual development, and the drafting, writing, and rewriting of manuscript content. Gemini (Google) was used to evaluate manuscript organization, conceptual consistency, internal coherence, clarity of argument, and overall scholarly presentation. Grammarly (Superhuman) was used for proofreading, grammatical correction, language refinement, and editorial review. All AI-assisted contributions were subject to author review and editorial judgment. AI-generated content was not treated as independent empirical evidence, a substitute for source verification, or scholarly validation. The author retains full responsibility for the manuscript's intellectual contributions, clinical interpretations, factual accuracy, scholarly claims, references, conclusions, and final published content. The use of artificial intelligence does not constitute independent peer review.

AI Tools Referenced

Google. (2026). Gemini [Large language model]. https://gemini.google.com/

OpenAI. (2026). ChatGPT [Large language model]. https://chatgpt.com/

Superhuman. (2026). Grammarly [AI-powered writing assistance software]. https://www.grammarly.com/