METHODS, EVIDENCE, AND LIMITS
Science & limits
Northbound uses behavioral research as a set of product constraints: define a bounded result, choose one next step, check what happened, and adjust from evidence. It is not a diagnostic system and does not claim that a method will work for every person or every situation.
What “science” means here
The science page is not a promise that Northbound can identify your psychology. It explains why the product asks for a result, constraints, evidence, review points, and a safe stopping rule before it suggests another step.
Research-informed methods are used as design inputs. Your actual Journey remains adaptive: the next step can shrink, move, pause, stop, or change direction when new evidence shows that the original plan no longer fits.
Why money examples appear
Early Northbound scenarios used financial situations because they make the product constraints visible: uncertainty, avoidance, hard limits, other people, dates, and consequences all show up quickly. Those examples are stress tests for the Journey system, not a claim that Northbound is a financial adviser.
In money-related Journeys, Northbound can help bound the decision, gather facts, choose one reviewable next step, and route out when a regulated or high-risk issue needs appropriate help. It must not recommend financial products, investments, debt strategies, legal actions, or choices that affect essential needs.
How scenarios are used
Northbound does not treat a scenario as a fixed category like money, work, health, or learning. A scenario is useful only when it describes the current situation, the result being tested, the constraint that matters now, and the kind of help that could make the next step safer or more realistic.
That is why the same template can lead to different next steps. One person may need to gather facts, another may need to reduce scope, another may need to wait for another person, and another may need to pause or change the Journey outcome. The public Journey topic is an entry point; the private Journey evidence determines the next decision.
From method to product decision
Behavioral methods such as action planning, implementation intentions, problem solving, reducing scope, environmental repair, social support, and Journey review are treated as possible tools. They do not run the product by themselves.
The product decision stays explicit: continue, shrink the step, replace it, pause, stop, change the Journey outcome, ask Northbound, or bring in a Companion. Wayfinder may suggest, summarize, and explain tradeoffs, but meaningful Journey changes require the participant's confirmation.
What Northbound does not do
Northbound does not diagnose motivation, personality, discipline, burnout, ADHD, anxiety, or any other condition. It does not treat a missed step as failure, weakness, or proof that you lack commitment. It also does not use public examples or template categories to infer hidden traits about you.
Order of checks
Safety, health and basic functioning, objective constraints, and privacy consent come before behavioral explanations. A method is used only for the current Journey decision, not for a personality type.
- Is there a safety, crisis, abuse, legal, medical, or regulated financial issue?
- Is the result still wanted and bounded?
- Is the next step actually possible with the current time, access, and energy?
- What evidence would show whether the step worked or needs to change?
- What is the smallest useful action or review point now?
How AI is evaluated
Wayfinder responses are tested against product policy, not against a claim of psychological validity. Automated evals check that the AI keeps the selected language, avoids diagnosis and guilt, preserves privacy boundaries, refuses hidden labels, respects safety routes, and proposes at most one primary action.
Synthetic evals help catch regressions. They do not prove that any particular Journey will produce a result for a particular person.
Method reference
The point of this section is fit, limits, and safe use. Each method is an option for a specific situation, not a prescription. Methods are selected only when the Journey evidence supports them.
Implementation intentions
Connect a recognizable situation to one feasible response. Northbound uses this only when it helps shape the next step.
Evidence status: Moderate, context-dependent. Effects vary with result quality, feasibility, and whether a stable cue exists.
Best fit: A chosen North, a repeated cue, and a small action that is already within the person’s control.
Poor fit: An unclear North, missing access, unsafe conditions, severe exhaustion, or a one-off decision without a repeatable cue.
Safe version: Choose one low-burden cue and action, then test it once. Leave the If/Then field empty when no reliable cue exists.
Stop or replace when: Replace the next step if the cue does not occur, the action is repeatedly infeasible, or the plan increases risk or strain.
Example: If I close my first work call on Tuesday, I will open the draft and edit one paragraph; after ten minutes I may stop.
References: Gollwitzer (1999), American Psychologist · Gollwitzer & Sheeran (2006), Advances in Experimental Social Psychology
Reduce scope
Make the next test smaller without claiming that smaller is always the correct Journey decision.
Evidence status: Context-dependent product heuristic. It is useful when burden is the limiting condition, but it is not a validated psychological scale or universal treatment.
Best fit: The North still matters, but the current Journey or action exceeds available time, energy, money, or attention.
Poor fit: The result is unsafe, externally imposed, no longer valuable, or blocked by an access constraint that a smaller action cannot change.
Safe version: Cut only one dimension: duration, frequency, or output, and preserve an explicit option to pause or reconsider the Journey.
Stop or replace when: Stop shrinking the task if the remaining action no longer tests the North or if repeated reduction hides a decision to pause or stop.
Example: Replace a five-session week with one 20-minute trial, then review whether the work still carries future value.
Environment repair
Change an external condition that makes the intended action unnecessarily difficult.
Evidence status: Context-dependent. Environmental cues and friction can influence behavior, but the relevant condition must be observed rather than inferred from a public problem label.
Best fit: There is direct evidence that access, setup, interruption, visibility, or another modifiable condition affects the action.
Poor fit: The constraint is structural or unsafe to change alone, or the user has not identified which environmental condition matters.
Safe version: Change one reversible condition while keeping the action and review point stable.
Stop or replace when: Undo or replace the change if it creates privacy, safety, relationship, financial, or accessibility costs, or produces no useful difference.
Example: For one session, put the required file on the home screen and silence only nonessential notifications; compare setup time afterward.
References: Wood & Runger (2016), Annual Review of Psychology
Small reversible experiment
Gather information before making a larger commitment. The experiment tests an assumption, not the person.
Evidence status: Information-gathering heuristic. Its value is data gain and reversibility; a favorable result is not proof that continuing is always correct.
Best fit: Uncertainty can be reduced through a bounded, low-cost action with a clear observation and no unacceptable downside.
Poor fit: Safety, health, legal, coercion, financial-dependence, or irreversible consequences dominate the situation.
Safe version: State one question, one reversible action, what will be observed, and a maximum cost or duration before starting.
Stop or replace when: Stop immediately at the predefined safety, burden, cost, or consent boundary; do not enlarge the test merely to rescue the hypothesis.
Example: Before paying for a course, complete one free sample lesson and note whether the format fits the available time and access needs.
Journey review
Separate past investment from future value when deciding whether to continue, pause, stop, or change the Journey outcome.
Evidence status: Mixed and context-dependent decision support. Research on sunk costs describes a common influence on decisions; it does not determine the correct choice for an individual.
Best fit: The participant is explicitly reconsidering a Journey and can examine future value, future cost, obligations, and transferable learning.
Poor fit: The decision is urgent or unsafe, key facts are missing, or another person controls the consequences and appropriate support is required.
Safe version: List future value, future cost, real obligations, and what remains useful if the current Journey closes. Choose only a reversible next decision when data are incomplete.
Stop or replace when: Pause the review if it becomes coercive, repeatedly rehearses past losses without new data, or requires professional legal, medical, safety, or financial guidance.
Example: Compare the next three months of cost and value under continue, pause, and stop; treat the years already spent as learning, not a command to continue.
References: Arkes & Blumer (1985), Organizational Behavior and Human Decision Processes
Review method fit, safe versions, stop criteria, and references where available above.
Limits of generalization
General research and public examples cannot guarantee an individual result. Synthetic evaluations test product policy and regression behavior; they do not establish psychological validity. Safety, medical, legal, violence, and financial-dependence situations require appropriate professional or local support.
Northbound is most useful when the situation can be handled as a bounded Journey: one result, one current step, explicit constraints, and a reviewable next decision. When those conditions are not present, the safer answer may be to pause, seek appropriate help, or not use Northbound for that decision.
Start from Explore Journeys if you want to see the current product shape.