Neuromarketing in brand perception research

Practical conclusion: if the task is to understand what a consumer will say to a salesperson or in a review, a survey is sufficient. If the task is to understand what a brand will really "grab" at the moment of choice on the shelf, where the decision is made in fractions of a second, physiological data are needed. This should be used not instead of, but together with: surveys provide language and context, neurodata — speed and honesty of reaction.

At what stage of the brain distinguishes a "own" brand from a foreign one

It was long believed that the brain first recognizes an object, then extracts brand memories, and then forms an evaluation — a linear decision-making pipeline. fMRI data using explainable AI (xAI) 2024 year contradict this model. The distinction between preferred and indifferent brands is already recorded at the stage of early visual processing — in the cuneal cortex and lateral occipital cortex, that is, literally in the first milliseconds of viewing a logo or packaging, even before the areas responsible for memory and evaluation become active.

This means that a brand is not "evaluated" by the brain as a separate step after recognition — it is tagged ("tagging") right at the entry point, in parallel with recognizing shape and color. Hence, a practical implication for design: the visual code of a brand does not work as a hint for a subsequent conclusion, but as a standalone trigger that either activates within the first fraction of a second or does not. Redesigning a logo "so that its meaning can be revealed later" is a strategy that contradicts how perception actually works.

N400 and other markers indicating that a brand "hooked" into memory

Among many EEG indicators, there is one especially valuable for assessing brand perception — the N400 component, a negative potential deviation occurring roughly 400 milliseconds after the stimulus. The weaker the semantic link between two concepts, the higher the N400 amplitude. Simply put: if the brain expects one word but receives another unrelated word, the amplitude increases.

In a study using semantic priming, Netflix and a fictional competitor were compared through associations with words like "television" and "relaxation." Surveys yielded the same result for both brands — respondents equated them equally with screen-based leisure. N400-analysis showed the opposite: the associative strength differed significantly between the brands, although this difference was not detected by the verbal instrument. The implication for practice: N400 is useful where one needs to check not the declared, but the actually embedded in memory association — for example, when selecting a slogan or positioning phrase that should "fit" the existing brand image, rather than create it from scratch.

In addition to N400, when analyzing a brand, one usually also looks at:

• the ratio of beta and alpha rhythms as an index of attention engagement;
• the skin conductance response (GSR) as a marker of emotional arousal without indicating its valence;
• eye movements and fixation time as indicators of visual capture, but not sympathy.

Hidden associations that a person is not aware of

The most inconvenient fact for marketers from neuroresearch — the discrepancy between physiology and self-report at the level of personality traits. People with high conscientiousness physiologically avoided brands with archetypes of "sincerity" and "brutality", although this did not manifest in questionnaires — they honestly answered that the brands were neutral or even sympathetic to them. The body reacted with rejection, while the language expressed neutrality.

This is exactly the zone for which neuromarketing exists not as a trendy add-on, but as the only available tool: hidden rejection or attraction that a person does not formulate because they are not aware of it. For positioning, this is critical when choosing a brand archetype for a specific segment — a segment with high conscientiousness cannot be convinced through "sincerity" in words if physiology signals the opposite well before purchase.

How much we can trust laboratory data

This is where the main gap lies between what is published in case studies and what actually happens on the real market. Machine learning models trained to recognize brand preference based on neurodata often achieve accuracy of 54–56% in laboratory conditions — slightly better than random guessing. Applying the independent component analysis (ICA) method to reduce data dimensionality can improve accuracy, but the improvements remain modest — progress is there, but it is too early to speak of a reliable forecast of individual choice.

At the same time, there is also a reverse, even more reassuring result: a study by the HSE University showed that the attention engagement index calculated based on EEG 25 people (the beta-to-alpha rhythm ratio) predicted the behavior of a real 291 301 consumer during a retailer's advertising campaign. The paradox is easily explained: individual prediction for one person is a task with enormous noise, whereas an aggregated signal for a group projected onto mass behavior proves to be stable. The limitations of the method must be honestly stated:

1. small samples (dozens, rarely hundreds of people) yield noisy individual predictions;
2. laboratory conditions (headset, chair, lack of real choice among shelves) distort the decision context;
3. ML models well interpret patterns at the group level, but poorly — at the individual level;
4. explainable AI (xAI) begins to show which specific brain regions contribute to the model's decision, but this 'black box' disclosure does not yet equal clinical accuracy of prediction.

How to use neuromarketing without inflated expectations

It is not advisable to purchase neuromarketing research as a replacement for advertising testing on real audiences — the individual prediction accuracy does not allow for that. A reasonable strategy is to use neurodata where a group, not individual, signal is needed: comparing logo variants based on early visual attention capture, validating a slogan for real semantic linkage through N400, identifying hidden rejection of an archetype in the target segment. And always correlate these data with behavior at scale — as in the HSE case, where a small laboratory sample was validated on a real campaign. It is precisely the combination of 'small precise signal + large market validation' that turns neuromarketing from a beautiful technology into a working tool for brand research.