The advice most people receive is to be more confident, which is roughly as actionable as being told to be taller. Confidence arrives as a feeling, so it gets treated as something you should be able to summon: decide to feel capable, walk in differently, and the rest follows. When it doesn’t work, the failure gets read as a character problem rather than as a flaw in the instruction.
The more useful account is that confidence is mostly downstream. It’s a prediction your mind makes about whether you can handle something, and it’s assembled from evidence — what you’ve done, what you’ve watched others do, what you’ve been told, and what your body is doing while you decide. Change the evidence and the prediction moves. That reframe is what makes the problem tractable, and it’s also what makes it slower and less glamorous than the genre suggests.
Confidence is an output, not an input
The clearest research foundation here is Albert Bandura’s work on self-efficacy, developed from the late 1970s onward. Bandura’s term is deliberately narrower than confidence: self-efficacy is your belief that you can execute a particular behaviour in a particular situation. Not a general sense of being an impressive person. A specific forecast about a specific task.
That narrowing is the whole contribution. Self-efficacy turns out to be one of the better predictors of whether someone actually attempts a difficult thing, persists after early failure, and recovers from setbacks. And because it’s a belief about evidence, it can be moved by supplying evidence.
There’s a related finding worth holding alongside it. A large 2003 review by Roy Baumeister and colleagues examined whether high self-esteem causes better outcomes, and found the causal arrow mostly runs the other way: performing well raises the appraisal more reliably than the appraisal raises performance. The same logic applies to confidence. It’s more consequence than cause, which is inconvenient if you were hoping to start with the feeling.
The four sources, in order of strength
Bandura identified four inputs to self-efficacy, and the ordering matters more than the list.
Mastery experience is the strongest by a wide margin: having actually done something like this before, especially something you weren’t sure you could do. Nothing else comes close, which is why confidence built entirely out of encouragement tends to be brittle.
Vicarious experience comes second. Watching someone comparable to you succeed at the thing raises your own estimate, and the comparability is load-bearing — watching an obvious expert does much less than watching a peer. This is the one place where comparison works in your favour rather than against you, and it depends entirely on who you choose to watch.
Verbal persuasion is third and weaker than most encouragement assumes. Being told you can do it helps a little, mostly when the person saying it has credibility and specifics. Vague enthusiasm from someone who’d say it regardless does close to nothing.
Physiological and emotional state is fourth. A racing heart before a presentation gets read as evidence of unfitness, and that reading lowers the forecast. Research on arousal reappraisal, notably Jeremy Jamieson’s work published in 2012, suggests that relabelling those signals as the body mobilising rather than as a warning produces measurable improvements in how people respond under pressure. The arousal doesn’t have to go away. Its interpretation is what’s negotiable.
That clip is aimed at the discount — the reflex that explains away every piece of evidence in your favour before it can be counted.
There is no single confidence dial
Because self-efficacy is task-specific, the everyday phrase “low confidence” usually collapses several separate ledgers into one verdict. Someone can be entirely steady presenting to a room and unsteady asking for what they want in a relationship. Those are different track records. Transfer between them happens, but it’s partial and much weaker than the single-dial model implies.
This is practically useful. “I need more confidence” has no first step. “I have no evidence that I can hold a boundary when someone is disappointed in me” points directly at what to go and generate. Naming the domain converts an identity complaint into a task, which is the same move that makes limiting beliefs workable: the generalised verdict is the problem, and specificity is what dismantles it.
It also explains a common trap. People often try to build confidence in the domain that hurts least, because that’s where progress is comfortable, and then wonder why the area that actually bothers them hasn’t moved. The ledger you avoid is the ledger that stays empty.
What hasn’t held up
Two pieces of standard confidence advice deserve honest treatment, because the cluster’s credibility depends on saying so.
Power posing is the better-known case. A 2010 paper reported that expansive postures raised testosterone, lowered cortisol, and increased risk tolerance. A larger 2015 replication by Ranehill and colleagues found the self-reported feeling of power survived but the hormonal and behavioural effects did not, and one of the original authors subsequently stated publicly that she no longer believed the effect was real. Standing differently may change how you feel for a minute. Treat it as that, not as a mechanism.
Broad positive self-statements are the second case. Joanne Wood’s 2009 work found that repeating lines like “I am a lovable person” left people with low self-esteem feeling measurably worse, because the statement created contrast with what they actually believed instead of overriding it. The instruction to affirm yourself into confidence fails hardest for exactly the people reaching for it. The power of positive thinking covers where the wider optimism literature does and doesn’t deliver.
Where affirmations actually fit
If confidence follows evidence, language can’t manufacture it. What language can do is protect the interpretation, and that turns out to matter a great deal, because evidence is not self-interpreting.
The same completed project reads as I pulled that off or as I got lucky and they’ll notice eventually, and the reading determines whether it enters the ledger at all. People with a long habit of discounting can accumulate a decade of mastery experience without the forecast ever updating. The bottleneck isn’t the evidence. It’s the accounting.
That’s the honest job description for confidence affirmations: maintenance of a fair reading rather than the manufacture of a feeling. Which is also why they need to stay in the almost-believe range and stay tied to something specific. “I handled the part of that I was dreading” is a claim your mind can check. “I am unstoppable” is a claim it will reject, and the rejection costs you.
Two adjacent practices carry more of the weight than people expect. Self-compassion determines what happens on the days the evidence goes the wrong way — whether a failure gets filed as one data point or as confirmation of a verdict. And growth mindset affirmations work on the assumption underneath the ledger: whether capability is fixed, in which case every setback is diagnostic, or built, in which case setbacks are just the cost of the build.
What to do this week
Pick one domain, stated narrowly enough to be a task rather than a trait. Then find the smallest action in it that you’re genuinely uncertain about — not the one that frightens you most, the next one up from what you already know you can do. Bandura’s own applied work made this point repeatedly: graded steps produce durable belief, and steps too large produce evidence of failure.
Do it once. Then write down what actually happened, in a sentence, before your memory begins editing. The writing is not optional, because the discount runs quietly and fast. Journaling is doing one narrow job here: keeping the record honest enough to be read later.
At night, resist the review. Substitute one line for it, something in the shape of there is more evidence than I have been counting, and let the next step happen in the morning, when it’s actually available to you.
Personalized confidence sessions with Murmora
Murmora is built around the accounting problem rather than the pep talk. You tell it the specific domain where the forecast is lowest — the negotiation, the boundary you don’t hold, the work you’re competent at and don’t believe you’re competent at — and it generates affirmations calibrated for that situation, in the almost-believe range, paced for the sleep-onset window in one of the guide voices. The point isn’t to talk you into a feeling. It’s to make sure the week’s actual evidence gets recorded at its real weight rather than discounted at 1 a.m.