Analysis of over 800 million tweets reveals how our thought patterns shift throughout the day

Our minds follow different patterns of thought throughout the day, social media analysis reveals.


Image via Maxpixel.

How does one glean insight into the human mind? One method is to look at tweets. Many tweets. Some 800-million tweets, judging by a novel study. The paper, published by University of Bristol researchers studied thinking behaviors by analyzing over seven billion words tweeted by Britons throughout the day over the past four years — and report that two main factors influence how we think throughout the day.

Thought swings

“The analysis of media content, when done correctly, can reveal useful information for both social and biological sciences,” said Nello Cristianini, Professor of Artificial Intelligence and lead researcher. “We are still trying to learn how to make the most of it.”

The team of researchers, with a strong background in both artificial intelligence (AI) and medicine, used AI software to analyze aggregated, anonymized UK Twitter content to understand how our minds work. The material was sampled every hour over the course of four years across 54 of the UK’s largest cities.

The researchers tracked the use of specific words, associated with 74 psychometric indicators, across the sample — which they then used to interpret the underlying thinking style. The results suggest that our thinking patterns change throughout the day, and follow a roughly 24-hour cycle.

Although they tracked 73 different psychometric qualities, the team found that it all boiled down to two independent factors that explain most of the variation seen in the dataset.

The first pattern of thought, the team reports, is a more analytically-inclined one. It seems to peak at around 5 to 6 am. Tweets sent out around this hour used words and an overall language style previously shown to correlate with more logical patterns of thought. They included a high ratio of nouns, articles, and prepositions, which the team notes have previously been linked to intelligence, academic performance, and education.

During these hours, people also showed increased concern with achievements and power.

In the evenings and during the night, however, the pattern flips. It becomes more emotional and takes on existential tones. It’s a more impulsive, social, and emotionally-heavy mode, and its expression peaked at around 3 to 4 am, the team reports. The algorithm the team employed found that during this interval there was heavy language correlated with existential concerns — but negatively correlated with expression of positive emotions.

The team notes that these shifts also occur during times associated with major changes in neural activity and hormonal levels — which would suggest that they’re tied to the workings of our circadian clock. Finally, they report that a user’s cognitive and emotional states could be reliably predicted over a 24-hour period.

The paper is awaiting publishing in the journal PLOS ONE. Materials via University of Bristol.

Leave a Reply

Your email address will not be published. Required fields are marked *