More Hypothesis Templates

Kim Kastens
Author Profile
published Aug 12, 2009

In an earlier post, I introduced the idea of "hypothesis templates," a device intended to help students learn to generate plausible hypotheses from spatial information. I'm still struggling with the question of how do people, either experienced geoscientists or beginning students, generate meaning from information about the position, configuration, trajectory, orientation or shape of objects or phenomena in the real world. Over the last few days, my collaborators and I have been going through student products from the Lamont Data Puzzle Project, a curriculum development project. Whenever we instructed the students to "Suggest a hypothesis to explain [the observations they had just made from data]," many students struggled and came up with little to nothing.

So today, I offer hypothesis templates for two additional common types of observations.

First, for the situation in which two phenomena, call them A and B, are closely associated in space and time. One issue is how to determine whether A and B are, in fact, "closely associated" at a statistically significant level. I'm not too worried about that at the moment; for the sake of discussion let's say that the spatial co-occurrence is compelling to the naked eye on a data visualization. Once we are in agreement that A and B co-occur, how can we think methodically about possible causes for this co-occurrence?

Generally speaking, the possibilities are that A causes B, B causes A, or some other factor or process causes both. Or, of course, the association may be pure coincidence and have no causal or process significance at all.

Hypothesis Template Table for Co-occurence

To take a geological example, if earthquakes and volcanoes are observed to co-occur, the possibilities are that earthquakes cause the vulcanism, or vulcanism causes the earthquakes, or that a third factor causes both. In this example, all three possibilities are plausible: earthquakes can open up a pathway for magma; movement of magma through subterranean plumbing systems can cause tremors; and on a mid-ocean ridge both vulcanism and seismicity are caused by the motion of the upwelling and diverging limbs of a mantle convection cell.

Photograph of Halite Crystals
Halite (NaCl) crystals. (image info)
Another type of spatial observation that calls for explanation is that an object of nature or group of objects of nature have a distinctive shape.

It takes a certain level of insight to even realize that shape is something that can be explained, explained by natural processes, rather than just accepted as the way things are. For example, the planar surfaces and right angles of these halite crystals reflect the regular lattice of Na and Cl atoms in the crystal structure--or stated more generally, the external form reflects the internal structure.

A suite of hypothesis templates for explaining shape:

Hypothesis Template Table for Shape

More Hypothesis Templates --Discussion  

This post was edited by Phuoc Huynh on Dec, 2016
NOVA Geoblog has a nice discussion with illustration of how to help geology students think about two different ways that something can get to be spherical: an oolite gets round by accreting on all of its sides as it rolls around and experiences chemical precipitation, but a quartz sand grain gets round by being abraded on all of its sides as it rolls around and experiences physical weathering. The observable result (a spherical grain) is similar, but the processes is different.

Callan's well- stated and well-illustrated example with the oolites and the clastic grain could be the basis for a new "hypothesis template."

If you observe something spherical or near spherical, possible causal hypotheses are that:
* it got that way by building up in layers, building up evenly on all sides (e.g. hailstone, oolite.)
* it got that way by starting out irregular in shape and had its corners and protrusions abraded off (e.g. quartz sand grain.)
* it formed under conditions where the internal gravitational attraction of the body itself was more than the gravitational attraction of external objects (e.g. a planet, ball bearings made in the space shuttle


Share edittextuser=75 post_id=3639 initial_post_id=0 thread_id=1243

In follow-up to my previous comment about a hypothesis template for spherical objects, the Sciences Times today has an article about spherical objects in nature and art: "The Circular Logic of the Universe," by Natalie Angier (


Share edittextuser=75 post_id=3649 initial_post_id=0 thread_id=1243

The New York Times has an intriguing article about the scientific controversy over the origin of the so-called "fairy circles" found in desert areas of Africa. These are circular areas that completely lack plants, surrounded by areas covered with grasses. Judging from the photos in the article, these things are about 20 feet in diameter, and there are hundreds or thousands of them. Hypotheses on the table include water competition, termites, and poison. The Nov 1, 2022 article reports on a new study that favors the water-competition hypothesis and weakens the termite hypothesis. In the context of "hypothesis templates," this controversy reaffirms that Earth and environmental scientists do spend considerable time, money, and ingenuity on trying to explain the causal mechanisms of distinctive geometrical and spatial features in nature.

"In hunt to solve 'fairy circle' mystery, one suspect is dismissed" by Rachel Nuwer


Share edittextuser=75 post_id=47544 initial_post_id=0 thread_id=1243

This post is now 14 years old, but I still find the idea of hypothesis templates useful as an entry point into thinking about what kind of causal mechanism could underlie a pattern observed in real world data. I would like to add one more hypothesis template to the "co-occurence in space of phenomenon A and B." Such a pattern of co-occurence could also be caused by a reinforcing feedback loop, in which an increase in A leads to an increase in B, and an increase in B also leads to an increase in A.

This line of thought was inspired by reading Ogorevc et al (2020) Social feedback loop in organic food purchase decision-making at: Those researchers applied a "spatial lag model" to a large international survey dataset and documented a pattern in which the more frequently the people "close" to an individual "make a special effort to buy fruit and vegetables grown without pesticides or chemicals" then the more frequently that individual will also make such a special effort. They interpreted this pattern as resulting from a reinforcing feedback loop working as follows: as the number of individuals doing behavior X increases, the social norm favoring doing X ramps up, and as the social norm ramps up, individuals will be incentivized to do X to fit in with the group, and the number of individuals doing X will also increase. The paper discusses implications for how to structure efforts to encourage purchase of organic food and other sustainability behaviors by leveraging such social feedback loops.

Interestingly, these researchers don't use the terms "close to" and "spatial" the way most geoscientists would, to refer to physical space. Instead, they define "neighbors" as individuals who reside in the same country, and have similar age, and years of schooling. They make a "dissimilarity matrix" based on gender, age, and years of school to quantify "distance." Then they quantify how much pro-organic-food social norm an individual would be exposed to by heavily weighting the survey answers of people "close to" that individual on this dimension of social distance.

In most data-based papers I have read about feedback loops, the researchers infer the possible existence of a reinforcing loop by looking for evidence of exponential growth or decay in behavior over time rather than by looking for co-occurence in "space." This is a different strategy for seeking answers to the eternal question of "what underlying mechanism could be causing this pattern that I am seeing in the world?"


Share edittextuser=75 post_id=47579 initial_post_id=0 thread_id=1243

Join the Discussion

Log in to reply