Doomscrolling

Doomscrolling refers to the compulsive consumption of negative news and distressing content, typically via social media or news feeds, despite the behaviour producing a measurable deterioration in mood. The person engaged in it is usually aware, in real time, that it is making them feel worse. They continue anyway. This is the feature that makes doomscrolling distinct from ordinary news consumption and the one that demands explanation: it is not hedonistic behaviour, it is not producing pleasure, and yet it persists.

The explanation begins with negativity bias, one of the most well-documented findings in cognitive psychology. The human brain does not process positive and negative information symmetrically. Negative stimuli receive more attentional resources, are processed more thoroughly, and are retained more durably than positive stimuli of equivalent intensity. This asymmetry is not pathological — it is the output of an evolutionary system in which threats required faster, more sustained responses than opportunities. An ancestor who over-attended to danger and missed a meal survived. One who under-attended to danger and missed a predator did not.

The problem is that this system was calibrated for an environment in which threats were local, legible, and finite. You could assess the threat, respond to it, and return to baseline. A negative information environment had natural limits. The modern feed has none. It is infinite, algorithmically curated, and optimised — deliberately — for emotional activation. Negative content consistently outperforms positive content on engagement metrics across platforms. Platforms respond to this signal by surfacing more of it. The feed becomes a threat-detection system operating in an environment where threats never resolve and never run out.

What keeps the loop running is not pleasure but urgency. This is the mechanism that distinguishes doomscrolling from other forms of compulsive digital behaviour. Pull-to-refresh and like counts exploit dopaminergic anticipation — the pleasurable possibility of reward. Doomscrolling exploits a different circuit: the vigilance system, which produces a sensation closer to anxiety than excitement, and which does not satiate the way reward-seeking does. The brain monitoring for threat does not conclude that it has seen enough bad news. It concludes that it has not yet seen all of it.

The feed architecture accelerates this. Infinite scroll was designed to remove the natural stopping cues that pagination provided. The bottom of a page is a decision point — a moment at which you must actively choose to continue. The feed removes this friction entirely. Autoplay, in the video context, does the same. The absence of stopping cues interacts with an already-activated vigilance system to produce sessions that extend far beyond any considered intention.

Research conducted during periods of acute collective stress — major elections, the early months of the COVID-19 pandemic, significant geopolitical crises — has documented the relationship between heavy news consumption and outcomes including elevated cortisol, disrupted sleep, increased anxiety, and impaired concentration. A study published in the journal Health Communication found that those who reported higher rates of "severely problematic" news consumption also reported significantly worse mental and physical health. The direction of causation likely runs both ways: anxiety increases news-checking, and news-checking increases anxiety.

The practical implication follows the same logic that applies to other engineered compulsive behaviours: willpower-based interventions operate too slowly for a system running faster than conscious deliberation. You cannot reason with an activated vigilance circuit in the moment of activation. Effective interventions must be structural: scheduled rather than on-demand news consumption, the removal of news and social media apps from the primary phone screen, the use of grayscale display modes that reduce the visual activation value of feeds, and the deliberate introduction of stopping cues — a timer, a physical location rule, a device boundary — that the architecture itself has removed.

Naming the behaviour is not incidental. Doomscrolling entered widespread usage around 2020, and the speed of its adoption reflects something useful: it gave people a concept with which to notice and describe what was happening to them. Concepts precede agency. You cannot decide to stop doing something you do not yet recognise as a thing you are doing.