Online harassment has been regarded as an inevitable feature of the internet since the day the internet began, a product of bad actors, mismanaged platforms, or a breakdown in civility as a whole. In investigative and legal contexts, it is often treated as a secondary concern, noise surrounding “real” crimes, or behaviour that is difficult to quantify and therefore difficult to address.

This framing is understandable. Harassment is messy. It crosses platforms, jurisdictions, and legal definitions. It often involves large numbers of participants acting in ways that appear irrational, disproportionate, or incoherent. But after years working as a digital investigator, with an educational background in behavioural science, I assert that the problem is not that harassment is inexplicable. It’s that we have been forced to engage with it through the wrong lens, and in a reactionary manner.

Harassment is not inexplicable. It is predictable once we understand the behavioural forces that shape it.

When we apply behavioural science to online harassment, drawing on established research in psychology, neurochemistry, and group dynamics, patterns begin to emerge. These patterns do not excuse harm. But they do explain it. And that understanding matters, because clearer explanations lead to clearer investigations. The person behind the behaviour is the perpetrator, and keeping that personhood in mind, focusing on the “why”s behind the action, leads to better investigative results and, eventually perhaps better mitigation tactics.

What We Often Get Wrong About Online Harassment

Much of the public conversation around online abuse focuses on platforms: moderation failures, algorithmic amplification, or insufficient reporting mechanisms. These factors matter, and they deserve scrutiny. But platform-centric explanations tend to flatten human behaviour into technical problems, obscuring the motivations and dynamics that actually drive harassment.

Another common framing reduces harassment to individual pathology: trolls, extremists, or “bad people” behaving badly. This too is incomplete. Many harassment campaigns are not driven by a single malicious actor, but by large groups of ordinary users participating in collective aggression, sometimes without fully understanding their own role in it.

Most large-scale harassment is not driven by a single actor, it is driven by group behaviour.

Investigators see this disconnect constantly. Victims experience overwhelming, sustained harm, while individual perpetrators often insist they were “just joking,” “piling on,” or “doing what everyone else was doing.” Without a framework to explain how normal people become part of harmful group behaviour, these cases are difficult to assess, triage, prosecute, and prevent.

How Behavioural Science Explains Online Harassment

Decades of research on crowd behaviour, anonymity, and social identity offer powerful tools for understanding online harassment.

Anonymity, for example, does not simply make people reckless. Research shows that anonymity shifts individuals away from personal identity and toward group identity. People become less guided by their own norms and more guided by the perceived norms of the group they identify with. In cooperative environments, this can increase prosocial behaviour. In hostile or toxic groups, it can dramatically increase aggression.

This is reinforced by what psychologists call the online disinhibition effect. Online spaces remove many of the social cues, such as facial expressions, tone, and immediate feedback, that normally regulate behaviour. Due to the asynchronous, abstract nature of online communication, a psychological distancing occurs. People are more likely to say or do things online that they would never do face to face, not because they are inherently cruel, but because the environment dampens empathy and consequence.

Layered on top of this are neurochemical dynamics, particularly involving dopamine and serotonin. Social media platforms are engineered around rewards – likes, shares, replies – that deliver dopamine hits. Dopamine is not just a “feel good” chemical; it is a motivator. It reinforces behaviour, encouraging repetition.

Harassment, especially in the form of pile-ons or moral punishment, is often rewarded with attention, validation, and engagement. Research has also shown that perceived moral punishment activates the dopamine pathways in the brain.

Outrage is not just emotional—it is neurologically rewarding.

When these mechanisms operate together—anonymity, group identity, disinhibition, and reward reinforcement—harassment stops looking random. It becomes predictable.

From Explanation to Investigation

Why does this matter for investigators and legal practitioners?

Because intent, escalation, and proportionality all look different when viewed through a behavioural lens. Understanding group dynamics helps distinguish opportunistic pile-ons from coordinated campaigns. It clarifies escalation patterns and supports more accurate assessments of harm, including foreseeable offline consequences.

Digital harm does not stay digital.

Looking Ahead

This publication is not a final word. It is a foundation.

As technologies evolve, the human mechanisms driving online harassment remain remarkably consistent. If we want a healthier internet—and a justice system capable of responding to digital harm with clarity rather than confusion—we must be willing to examine how people actually behave online, not just how we wish they would. The behavioural science of online harassment gives investigators the clarity needed to respond with precision rather than confusion.

Understanding behaviour is not leniency. It is precision.

What This Means for Investigators

Behavioural insight supports operational clarity.

When harassment is viewed through a behavioural lens, investigators gain practical advantages:

  • Better triage:
    Distinguish spontaneous pile-ons from organized campaigns early.
  • Clearer intent analysis:
    Identify ringleaders, recruiters, and escalation drivers versus peripheral participants.
  • Stronger evidence narratives:
    Contextualize behaviour patterns across platforms instead of treating incidents in isolation.
  • Improved harm assessment:
    Link online conduct to foreseeable offline consequences.
  • More proportional outcomes:
    Support differentiated charging and sentencing recommendations based on role and behaviour, not just volume.

Behavioral understanding does not replace technical forensics or legal analysis; it strengthens them.

Read the Presentation How Harassment Shaped the Internet HERE

About the Author:
Maria Thomas is a digital investigator and researcher with the RISE Information Security Foundation. With a background in behavioural science, her work focuses on online harassment, networked aggression, and the intersection of human behaviour, technology, and justice.