The last 12,000 years have been much cooler than previously thought, according to research published Wednesday in the journal Nature. And in contrast, human-caused warming of the atmosphere is even more anomalous than we’d realized. That’s because the study’s authors have found a new way to estimate historical temperatures which they say filters out seasonal shifts that had made past millennia seem warmer than they really were.

The findings offer a possible solution to an outstanding riddle about the recent history of climate change. The problem, called the Holocene warming conundrum, is that previous reconstructions of the historical climate showed a warm period from 6,000 to 10,000 years ago, followed by a period of cooling. Climate models of the same period, however, suggest that the planet would have been warming steadily.

By fine-tuning how we interpret the physical evidence of the changing climate, explains Samantha Bova, a paleoclimate researcher at Rutgers University and one of the study authors, “[the data] do show a warming that’s highly consistent with what is predicted by climate models.”

The reinterpretation of climate history changes the context of warming caused by humans over the last 150 years. Instead of warming up after a 6,000-year cold snap, Bova says, we’re headed into uncharted territory: a world warmer than at any time since glaciers last covered Manhattan.

It’s not that the existing interpretation that shows warming 6,000 years ago is wrong, Bova cautions. It just appears to reflect changes in summer temperatures in the northern hemisphere, not annual temperatures. That’s important because seasonal variation is driven more by the relationship between the Earth and the sun, and can’t be compared to annual averages predicted by climate models.

During the early Holocene, summers in the northern hemisphere were hotter as the Earth moved in a more oval-shaped orbit, causing larger swings between hot and cold. But that didn’t mean that the average year was warmer—just that seasons were more variable.

“Even before the Industrial Revolution, greenhouse gases have been increasing since 6,000 years ago,” Bova says. But that increase is tiny compared to modern emissions. The concentration of carbon dioxide in the atmosphere increased by about 20 parts per million between roughly 4000 BCE and 1850 CE, she said. By comparison, it’s increased by about 100 parts per million just since 1950.

But the conundrum has been fuel for climate skeptics, she says.

The argument goes: how can the greenhouse effect be so important for global warming if greenhouse gases increased during the Holocene at the same time as global temperatures declined?

Solving the puzzle began in 2016, on a Rutgers-led expedition to the northern coast of Papua New Guinea. There, Bova and her colleagues collected sediment cores from the seafloor spanning hundreds of thousands of years of climate history.

Estimating seasonal patterns during the Holocene is complicated by changes in greenhouse gases and in the extent of the ice sheets, and so the climatologists instead looked back to the last interglacial period, which lasted from 128,000 to about 115,000 years ago.

That last warm snap was climatologically simpler, Bova explains. “Unlike the Holocene, greenhouse gases are pretty much flat across that period.” Ice sheets also retreated quickly, leaving relatively few forces on the global climate.

At the same time, the Earth’s orbit around the sun was more oval, and seasonal variation in sunlight in the tropics was double what it is today.

Because of that, says Bova, “We can attribute changes in temperature that we see in our records pretty much solely to changes in incoming [solar radiation].”

And in fact, the long term variations that they saw over the course of the last interglacial period lined up with  seasonal sunlight patterns caused by the changes in Earth’s orbit, indicating that the sediment record was reflecting more intense summers, not hotter overall years.

From there, it was a matter of calculating the impact of seasonal sunlight on ocean temperatures, and subtracting it out to get an annual estimate.

When they calculated average annual temperatures by applying the method to existing temperature records from the Holocene, they found a pattern of steady warming—exactly what climate models had predicted.

And because the models aligned so closely with their record, Bova says, the team could begin to attribute warming trends in their data to physical causes.

At first, warming over the course of the Holocene was driven by the slow retreat of the ice sheets. “You’re reducing the amount of shiny white stuff you have on the surface of the Earth,” Bova says, “and less radiation is reflected back to space,” so the planet warms. Then, about 6,000 years ago, rising greenhouse gas concentrations led to further warming.

Bova acknowledges that there are competing hypotheses for explaining the discrepancy between models and previous climate data. One other hypothesis, for instance, argues the current climate models don’t do enough to account for the increase in atmospheric dust during the Holocene, which would have a net cooling effect.

But she says that this study provides evidence that the issue is with the interpretation of underlying data, not the models.

“It had been suggested before that these records might be seasonally biased,” says Bova, “But there was never a great way of proving it.”