Cognitive Bias and Climate Change


Naomi Orestes is a professor at Harvard, and writes a monthly column for Scientific American. In a recent column, she noted that climatologists have consistently under-predicted the rate of climate change in the Arctic. In fact, it seems to be warming about four times as fast as most models have predicted. Some reports have it warming faster than any of the models have predicted.

There are a lot of reasons why the experts predicted low. Orestes suggests one of them may be that climatologists are gun shy of the noisy, extreme climate science deniers, and have unconsciously experienced cognitive bias as a result. Better to aim low than risk attack by aiming high.

WC pretends to no particular expertise in either climate science or cognitive bias, other than being modestly well read in both and having lived most of his life in the subarctic region.

But what’s striking about climate change in the arctic are the number of feedback loops that have been and may yet be triggered by even modest temperature changes.

Arctic sea ice, for example. The white of ice and snow on ice reflects heat back up into the atmosphere. Dark, open waters absorb heat instead. As sea ice melts earlier and stays melted longer, more of that dark, heat-absorbing water is available to absorb sunlight. Especially in the spring and early summer when, in prior decades, that warming sunlight was reflected away. It’s a feedback loop: more open water, more heat absorption, earlier melting of thinner sea ice. And so we have this:

Sea ice extent in the Arctic at the end of the summer melt season each September from 1979–2022, based on satellite observations. The total extent is the area of all pixels in the satellite image where the ice concentration is at least 15 percent. The amount of sea ice ice that survives the summer melt is declining rapidly. NOAA Climate.gov graph, based on extent data from the National Snow and Ice Data Center.

That’s a loss of 31,100 square miles—an area the size of South Carolina—per year. Far beyond what most climate models WC has seen predicted.

Another example is permafrost. The permanently frozen soils of the arctic and subarctic contain an estimated 1.5 billion metric tons of carbon. It’s preserved, and out of our atmosphere, for the same reason that food doesn’t spoil in our freezers. Frozen stuff doesn’t support bacteriological activity. But as those frozen soils thaw, bacteria have the opportunity to metabolize that carbon and excrete it as CO2. Food in our refrigerators, in contrast to our freezers, can spoil. Bacteria are slowed at fridge temperatures, but still active. They are already acting on the carbon where permafrost has melted. Those 1.5 billion metric tons of carbon represent more than twice the amount of CO2 we already have in our atmosphere. More warming, more permafrost melting, more bacteria, more CO2. And it’s getting perilously close to thawing at depth.

Permafrost temperature at borehole sites in interior Alaska, measured at the end of summer (approximately September of each year). Measurements at these sites were taken at depths between 49 and 85 feet. Data source: University of Alaska Fairbanks, 2021

Note these temperature measurements are deep down in the permafrost. Upper layers are warmer, sometimes much warmer.. There’s a lot of carbon closer to being under bacterial action above that 49 foot deep sensor.

Methane bubble trapped in ice, Lake Baikal, Siberia (Credit: Stanislav Tolstnev, Siberian Times)

When the bacteria work on the thawed, trapped plant material in a low oxygen environment, it’s even worse: the emissions are methane, not carbon dioxide, and methane is a much more powerful greenhouse gas.

Glaciers also reflect sunlight that would otherwise be absorbed by the darker ground. They serve as reservoirs of water and reservoirs of cold. Anyone who has been near an active glacier knows the air and vegetation around the ice is colder than the air and ground a mile away. Melting glaciers are another feedback loop. And the ice is melting rapidly. Scientists monitor a selected sample of glaciers across North America, the “Reference Glaciers,” keeping an eye on their mass balance, the ratio between accumulating snowfall and summer melt.

Cumulative change in mass balance of a set of “reference” glaciers worldwide beginning in 1956. The line on the upper graph represents the average of all the glaciers that were measured. Negative values indicate a net loss of ice and snow compared with the base year of 1956. For consistency, measurements are in meters of water equivalent, which represent changes in the average thickness of a glacier. The small chart below shows how many glaciers were measured in each year. Some glacier measurements have not yet been finalized for the last few years, hence the smaller number of sites.

The reference glaciers are shrinking, and, in a classic example of a feedback mechanism, the rate at which they are shrinking is increasing.

Feedback loops are notoriously difficult to predict. Often, they are exquisitely sensitive to very modest changes. WC suggests there’s a reason (besides cognitive bias) why climatologists have underestimated arctic and subarctic melting: an incomplete understanding of the feedback loops that are operating in those regions. If WC is right, it doesn’t make the rate of change any less scary or any less concerning; just a bit less inexplicable.