When you cite yourself, you might be doing your job right
Marvin Starominski-Uehara explores how a theory linked to termite behaviour challenges assumptions on self-citation and academic misconduct
Research integrity committees tend to frown upon self-citations. When a researcher cites their own work, colleagues and reviewers often read it as an attempt to inflate their profile, to game citation metrics. This concern is so great that it has been embedded in institutional check systems, with journals tracking self-citation rates and funders scrutinising authors whenever they appear.
I do not disagree with this default suspicion, but what if citing yourself has nothing to do with unscrupulous behaviour? What if it is a healthy practice in how a scientific field emerges?
In my study, published in Society, drawing from Pierre-Paul Grassé’s 1959 theory of stigmergy, I shared how termite colonies build elaborate nests. Termite workers respond to physical traces left by their own prior actions and the actions of others, and those traces guide what gets built next. The colony organises itself through accumulated signals.
Using this uncommon background, I asked whether academic knowledge grows the same way. If it does, then citing your own earlier work is not a failure to recognise the literature nor a shortcut for honest, organic credibility. It could be, instead, the trace a researcher leaves for readers and for their own thinking, a marker of where an idea has been and where it might go next.
Testing the assumption against data
To test this, I pulled together the most cited literature on human stigmergy itself. The search covered nine languages and turned up 1,289 entries, which I screened down to a sample of 42 papers meeting citation and quality thresholds. I coded each paper for its self-citation rate and cross-checked that against how often it went on to be cited by scholars with no authorship overlap.
The regression results ran counter to the standard suspicion. Self-references predicted future non-self-citations, meaning citations from unaffiliated authors, with a coefficient of 11.09 and a p-value of 0.010. This means that every additional self-reference in a paper was associated with eleven more citations from outside authors down the line. Non-self-references, however, references to other people’s work, showed no such relationship, and this pattern held up in further tests.
What I think self-referencing signals
I interpreted this finding through stigmergic logic. A termite returns to what it is building before moving elsewhere. That return reinforces a structure and signals to other workers where activity is concentrated.
When a researcher builds on their own prior findings, they are marking out a line of inquiry, showing that a claim was part of a developing body of original work, so citing their own work becomes the natural incremental next step, expected and justifiable, in developing an emerging field.
What is interesting to observe is that, just like termites’ collective behaviour, researchers operating in the same environment pick up on that signal. A paper acknowledged by peers for a visible trajectory of meaningful research, one an author keeps extending, reads not as an attempt to game anything but as a sustained effort that is worth building upon.
I must, however, be careful not to overstate this. The regression model explained only 19.3 percent of the variance in citation outcomes, which means that other factors such as journal prestige, author reputation and topic trends account for most of what drives a paper’s impact.
In addition, the sample is limited and the study field is quite specific, which limits how confidently the findings can be generalised. It would be great to see other researchers use such a framework to guide data collection and analysis, to counter, add nuance to or confirm these findings.
Why this changes how I see integrity policy
The findings suggest that a rigid threshold might mistake a legitimate signal for a red flag. An author building a coherent, cumulative line of innovative work should be expected to generate a higher self-reference rate. Punishing the rate risks punishing genuine, consistent effort and coherent, valuable scientific insight.
None of this means I think self-citation abuse does not exist. Excessive self-citation and reciprocal citation practices, for instance, are real problems that integrity offices are right to watch for. What my study adds is a caution against treating the mere presence of self-reference as evidence of that behaviour.
The stigmergic frame draws a line between a scholar marking a legitimate trajectory and a scholar manufacturing the appearance of one. Perhaps the marks scholars leave on their own path deserve more trust than integrity policy currently gives them.
Marvin Starominski-Uehara teaches at Temple University Japan and researches emerging intelligence in complex systems. He is an interdisciplinary scholar and the principal investigator of the Stigmergy Network Theory project