The Hidden Trade-Off Between Money and Relationships in Job Decisions
A job offer can look generous on paper but still ask for something that salary alone cannot replace. This tension lies at the heart of a groundbreaking 2025 study that bridges three usually separate areas of human experience: self-reported happiness, the concepts that come spontaneously to mind, and imagined decisions weighing close relationships against financial opportunity.
Drawing from an online sample in the United States and United Kingdom, the study found that individuals with higher subjective wellbeing generated chains of words semantically closer to “friend.” Interestingly, their responses did not show a similarly consistent relationship with “money.” A parallel dataset from South Korea replicated the friendship-related pattern using a related task, underscoring a potential cross-cultural consistency.
Days later, participants from the English-speaking group faced a hypothetical scenario: a lucrative job offer requiring relocation abroad, which would limit regular access to close relationships. Those whose earlier word associations were closer to “friend” demanded a higher salary bump before accepting the role—revealing that what spontaneously came to mind predicted what people later stated they valued most.
While compelling, these results come with important caveats: the semantic chains represent accessible concepts under time pressure rather than a full picture of mind-wandering, and the job choice was hypothetical rather than real.
“Spontaneous” Associations Under Timed Conditions
Published in Communications Psychology in November 2025, the study was led by Won-Gyo Shin and colleagues and involved 210 native English speakers aged 18 to 35 from the US and UK, recruited through Prolific in September 2023. Participants reported no current mental health issues, with a demographic majority identifying as White and a mean age just over 28.
The researchers first measured positive affect, negative affect, and life satisfaction through questionnaires, combining these into a standardized subjective wellbeing score. Then, participants completed the Free Association Semantic Task (FAST), a method designed to capture rapid, self-generated conceptual chains. Starting from three seed words—“key,” “friend,” and “money”—participants typed words or short phrases that came to mind within seven seconds per response. Each seed generated 30 responses, totaling 90 per participant.
It’s important to clarify that “spontaneous” here refers to the flow of associations under experimental constraints rather than an unstructured recording of everyday thoughts. Participants consciously responded to prompts, making this a structured but rapid glimpse into semantic accessibility.
Mapping Words to Meaning Using Language Models
To quantify thousands of responses such as “dinner,” “laughter,” and “support,” the team employed Word2vec—an advanced language model trained on Google News data. This tool represents words as vectors in a 300-dimensional semantic space, positioning words used in similar contexts nearer to each other.
Two main metrics were used: cosine similarity, which measures the angle between a participant’s response vector and the target concepts “friend” or “money,” and Word Mover’s Distance, estimating how far one set of words would need to “move” in semantic space to align with the target.
Additionally, researchers examined whether the chains converged progressively closer to “friend” or “money” as the task proceeded, rather than only calculating an average distance.
However, these models interpret language patterns rather than personal meanings. For example, the word “home” might evoke warmth for one participant and conflict for another. Thus, the study provides a map of what concepts are readily accessible semantically rather than a direct window into individual minds.
Positive Affect Drives the Friendship Connection
Participants with higher subjective wellbeing consistently generated responses semantically nearer to “friend,” showing greater cosine similarity and shorter Word Mover’s Distances. Their chains also tended to move closer to friendship-related concepts over time.
By contrast, money-related semantic patterns showed no significant or consistent link to wellbeing. This asymmetry suggests that friendship-related mental accessibility aligns more closely with feelings of happiness than money-related accessibility.
Digging deeper, positive affect—the frequent experience of positive emotions—was the strongest driver of this pattern, rather than life satisfaction or reduced negative affect. This distinction reflects broader psychological findings that pleasant feelings and life meaning represent different facets of wellbeing, as previously discussed by Silicon Canals in this analysis.
Notably, even in chains initiated by the word “money,” participants with higher positive affect showed a drift toward friend-related concepts, highlighting a social orientation that transcended the immediate seed word.
Cross-Cultural Replication in South Korea
To assess whether the friendship association extended beyond English speakers, the researchers analyzed a 2019–2020 South Korean dataset with 350 participants, mostly college-aged around 23 years old. Using a similar FAST task but with different seed words such as “family,” “mirror,” and “tear,” participants produced 40 responses per seed.
After translating Korean responses into English and applying the same Word2vec analysis, the study found that higher subjective wellbeing again correlated with semantic proximity and convergence toward “friend,” but not “money.” Positive affect remained the primary contributor to this effect.
This replication is meaningful, though not identical due to differences in language, age, seed words, and task design. Only the US/UK group faced the financial-versus-relationships dilemma, and cultural factors intertwined with other variables remain to be disentangled in future research.
Interestingly, the US/UK participants showed stronger money-related associations than the Korean group, a difference whose causes are yet to be explored.
Imagining the Cost of Relocation
Within three days of the initial task, 192 participants from the English-speaking sample completed a second online session. They read a scenario describing a job offer from a top-tier global company promising higher pay, bonuses, and financial benefits—but requiring an international move eight hours away from friends and family, with uncertain return possibilities.
Participants indicated the minimum salary increase (ranging from 20% to 200% in 20% steps) that would make them willing to accept the position, or if no increase would suffice.
Those whose earlier concept chains were semantically closer to “friend” required larger salary increases before agreeing to relocate. This relationship held after controlling for age, sex, education, socioeconomic status, and materialism.
The researchers validated the model using leave-one-out cross-validation to ensure robustness, though they caution that replication with a new sample and real-world decisions is necessary.
This hypothetical scenario simplified real-life complexities—such as family dynamics, visas, career stage, or hybrid work—that influence relocation decisions. The clean trade-off between money and relationships was deliberately isolated to reveal this underlying psychological pattern.
Interpreting Statistical Mediation and Causality
The study also explored a mediation pathway: higher subjective wellbeing predicted friend-related semantic similarity, which in turn predicted a higher salary threshold for relocation. While this statistical alignment is informative, it does not establish causality, especially since the mediator was not experimentally manipulated.
Close relationships could foster positive emotions and make social concepts more accessible. Conversely, positive emotions could broaden social attention and motivate investment in relationships. Personality traits, financial security, and past social experiences likely influence both pathways. These processes may reinforce each other over time.
Earlier research supports this complexity. For instance, a 2002 study of 222 university students found the happiest group to be highly social, with stronger romantic and other relationships, though no single factor fully explained happiness (source).
Since the 2025 study did not manipulate thoughts or track actual career decisions, and its main data collection was not preregistered, it cannot claim that thinking about friends increases happiness or that happiness directly causes people to reject money.
Valuing Friends Over Money—But Money Still Matters
The absence of a consistent money-related semantic association should not be interpreted as money being unimportant to wellbeing. Financial resources provide security, reduce stress, support loved ones, and make social time more feasible.
Indeed, the study found one exception: life satisfaction correlated positively with convergence toward “money” when money was the seed word. This finding aligns with the idea that financial resources may impact broader life evaluations more than everyday positive feelings.
Furthermore, “friend” served as a proxy for the social world, which also includes family, romantic partners, neighbors, and community—especially important in cultures where family ties are central.
Limitations include the narrow demographic of young adults without reported mental health issues, translation layers in the Korean sample, and potential cultural nuances lost in the English Word2vec model.
Importantly, the authors have shared their data and analysis code through the Open Science Framework, inviting transparency and further research.
In conclusion, the study does not claim that happy people disregard money. Rather, it shows that happier individuals—especially those experiencing more positive affect—navigate a semantic landscape more oriented toward friendship. When faced with a hypothetical choice, this landscape predicted how much financial incentive they would require to sacrifice close relationships.
Read more Here.
