Topicsshowing only Interpretation/Inference Show all Topics
I've been thinking a lot recently about how scientists and students make meaning from data, spurred in part by the Earth Cube education end-users workshop. Among other things, I've been trying to understand what kinds of deeply foundational understandings might be constructed by young children through unstructured observation using the human senses, and then later re-purposed as they begin to work with data.
Here is one candidate: Future data users need to understand that:
- events in the world leave traces, and
- by looking the traces, we can make inferences about the events.
Carol Cleland (2001, 2002) has written eloquently about how geologists do their science by examining and interpreting the traces left by the events of the past. However, it seems to me that at some level many (or maybe even all) data sets can be viewed as traces of events. Many of our scientific observation techniques are attempts to generate artificial traces (aka "inscriptions") for phenomena which do not leave natural traces. Think of tracks of drifting oceanographic buoys, or sonograms of bird calls.
It seems to me that this understanding--that events leave traces, and by looking at the traces we can make inferences about the events--is very deep seated in humans, and begins to develop very young. As an example around which to build up my thinking on this topic, I've been considering the case of a spilled glass of milk.
- The glass used to be vertical, and the milk used to be inside the glass.
- Something knocked the glass from the side.
- The glass rotated from vertical to horizontal, and the milk exited from the glass and spread across the table.
This mental model actually has quite a few sophisticated features, features in common with adult scientific models.
First of all, there is the concept of the active agent, the "something" that knocked the glass from the side. This agent has certain known attributes but is nonetheless not precisely specified. The active agent was certainly moving. The agent was most likely alive (a cat, a child), but could possibly have been inanimate (a ball, a big gust of wind). The ability to populate a mental model with an agent that has some known attributes or behaviors, but that is not individually specified, seems to me like an important pre-science interpretive skill, which could underlie the ability to entertain multiple working hypotheses in science.
Consider also the timing of the event. The child knows that the spilling event happened before he or she entered the room, because s/he doesn't remember seeing it happen. And s/he can infer that it didn't happen many days ago, because if that were so the milk would be dried up rather than fluid, and it would smell bad by now. Thus the timing of the event is bounded, but not specified. This is a common situation for geologists, who often can put upper and lower boundaries on an event in the geological record, but cannot pin it down to a specific date, as for example, when Bill Ryan and Walter Pitman were first trying to pin down the date when the Mediterranean waters spilled into the Black Sea.
In addition, this mental model involves some notion of normalcy, that it is normal for a glass to be vertical and for milk to be inside the glass. And some concept of fluids, to allow the notion that the milk used to be in a cylindrical shape but changed into a thin sheet when it was released from its confining container.
Further refinements are possible. The observer could infer the direction of motion of the active agent by the direction that the milk was spread out relative to the glass. And s/he could infer something about the vigor of the knock and the rotational velocity of the tipping glass by how elongate the spill was in the inferred direction of the impact.
Development of this ability could be researched by giving kids of various ages a drawing or photograph of the end state of a trace-producing event, asking them to draw a series of pictures showing how this scene got to be the way it is, and then explain their drawings to the experimenter. This same task could also be used as a learning activity, rather than a research task. I would speculate that younger kids would sketch just one working hypothesis, one sequence of events (for example, just the cat knocking over the glass) and would not allow for other possible working hypotheses. Eventually kids would get to an age were they could entertain multiple specific working hypotheses (maybe it was the cat, maybe it was younger brother), especially when questioned about whether there might be any other possibilities. Quite a bit later, I suppose, would come the ability to grasp and articulate the generalization that there was a knocking-over-agent with an unknown identity but certain knowable attributes.
- First, there was a blank sheet of paper.
- Then, someone drew the drawing.
- Then, someone ripped apart the paper.
Once again, we have the idea of an active agent, the one who drew the picture. It definitely was not the cat. It was probably a child, based on the character of the drawing. But we can't specify which child. Then an active agent ripped the paper. We can't tell if the ripping agent was the same or different from the drawing agent. The second agent could have been the cat. Or the drawing child. Or a different child. Again, we have the notion of active but unobserved agents, with some knowable attributes and other attributes that can be bounded but not pinpointed.
The mental model for the ripped paper also has a temporal constraint, but it is a different kind of temporal constraint from the bounding of time in the spilled milk case. In the case of the ripped paper, we know that the drawing was made before the paper was ripped, because the rip cuts across the drawing. This line of reasoning is identical to that which geologists use when they infer that a fault post-dates the deposition of strata which it offsets. There are also some more subtle sequencing inferences possible within the drawing: most likely the frame of the house was drawn before the chimney, doors and windows. But there are some temporal details that cannot be inferred from the trace: for example, we cannot infer whether the chimney was drawn before or after the door.
I don't think it is asking too much to expect that preschool aged children should be able to tackle these kinds of problems, either in a research setting or as learning activities. Gopnik and colleagues (Gopnik, 2012; Gopnik, et al. 2004) have shown that students in this age range have the ability to construct abstract, coherent, learned representations of causal relations among events and then use these representations to make causal predictions. I am asking whether students can make retro-dictions as well as pre-dictions.
- about events:
- Shipley, T. F. (2008). An Invitation to an Event. In T. F. Shipley & J. M. Zachs (Eds.), Understanding Events: From Perception to Action (pp. 3-30). Oxford: Oxford University Press.
- about events leave traces:
- Cleland, C. (2001). Historical science, experimental science, and the scientific method. Geology, 29, 987-990.
- Cleland, C. E. (2002). Methodological and epistemic differences between historical science and experimental science. Philosophy of Science, 69, 474-496.
- about kid's scientific thinking:
- Gopnik, A. (2012). Scientific thinking in young children, theoretical advances, empirical research, and policy implications. Science, 337, 1623-1627.
- Gopnick, A., Glymour, C., Sobel, D. M., Schulz, L. E., Kushnir, T., & Danks, D. (2004). A theory of causal learning in children: Causal maps and Bayes nets. Psychological Review, 111, 3.
I have a long-standing interest in the use of data in education, so I've been reading with interest several articles and a book concerned with the so-called "Fourth Paradigm" of science, in which insights are wrested from vast troves of existing data. The Fourth Paradigm is envisioned as a new method of pushing forward the frontiers of knowledge, enabled by new technologies for gathering, manipulating, analyzing and displaying data. The term seems to have originated with Jim Gray, a Technical Fellow and visionary at Microsoft's eScience group, who was lost at sea in 2007. The first three paradigms, in this view, would be empirical observation and experimentation, analytical or theoretical approaches, and computational science or simulation. Earth and Environmental Sciences are well represented in the book, with essays on data-rich ecological science, ocean science, and space science.
I am finding these readings very stimulating and worthwhile. But I question whether this way of making meaning from the complexity of nature is really so new. More
I'm thrilled to report that the book that grew out of the Synthesis project, the parent project of this blog, is now out: Earth & Mind II: A Synthesis of Research on Thinking and Learning in the Geosciences, Geological Society of America Special Publication 486, edited by Cathy Manduca and myself. It's available from the Geological Society of America bookstore
However, having shared my thrill at holding the book in my hands, I have to admit that there are some ideas in the book that I have already outgrown during the months that the book has been in production. More
Recently my "Teaching & Learning Concepts In Earth Sciences" students and I renovated one of my old data-using lab activities, from the days when I used to teach "Planet Earth" to non-science majors. The old version of the activity led students step-by-step through a series of manipulations of an on-line global data base, using a professional data visualization tool. The old directions provided a lot of scaffolding for how to make data displays of ocean salinity in and around the Mediterranean Sea, but little support for how to extract insights about earth processes from those displays. The new version assumes that students are already pretty adept at getting computer apps to do what they want, and refocuses the scaffolding on how to think like a geoscientist, how to think about the meaning of the data. More