I’ve been thinking a lot about the human tendency to categorize lately and the ways in which our methods of categorizing the world affect our ability to draw accurate maps of it, especially as it impacts science. I think it has come to the attention of a lot of people that the majority of the most contentious debates in the human sciences (such as psychology, anthropology, and biology) are specifically over the ways we classify things.In paleoanthropology, career-long disputes are formed over which category, if any, certain fossils fall into. In psychology, where I think the debate is luckily a little more empirical, disputes are formed over what the appropriate way to classify a personalities. In human genetics, the question of classification dangerously teeters between the two counter-notions that human races are not real but that humans can generally be split into five or six continental populations (if you so wish to classify them that way). In my own online disputes over sex concepts, I have belabored the point that there are probably several sex categories (gametic, chromosomal, genital, and gendered) rather than only one or simply none.
The way we classify things has serious implications for science. If we can’t agree what a thing is or if an errant phenomena even constitutes a thing, then the ability to have a discussion about the processes, relationships, and categories in thingspace are critically stunted. I strongly believe that a critical problem here is while disagreements between details are important, some of the disagreements in science aren’t even about details. In fact, a great number of these debates are caused by the fact that some people aren’t even seeing the world the same way as one another.
Ways of Seeing
During my first year of graduate school, one of my committee members, Mike Manson, who looks a bit like Noam Chomsky, introduced me to a dichotomy used by a subset of computer scientists to describe the ways people think about things, now referred to as the Mapper/Packer orientation. This viewpoint was first described in a programming blog from 1997 as follows1:
Mappers predominantly adopt the cognitive strategy of populating and integrating mental maps, then reading off the solution to any particular problem. They quickly find methods for achieving their objectives by consulting their maps.
Packers become adept at retaining large numbers or knowledge packets. Their singular objective is performing the `correct’ action. Strategies for resolving ‘hash collisions’, where more than one action might fit a circumstance are \ad hoc\.
I want to quickly clarify that I’m not an expert in two of the things I’m about to talk about. First, I’m not an expert in set theory, or the philosophy of this sort of thing. In this post I’m strictly talking about cognition, which is my second disclaimer: I’m not a trained cognitive scientist. Despite working on behavioral evolution, I am not familiar with much of the background literature in this somewhat amorphous and quickly evolving field. I bet other people have already talked about this, so I’m sorry for bulldozing any nuance you guys may have built here.
What I will say is that I think the computer programmers have grabbed onto something important. In a sense this binary division describes people who are actively able to integrate novel or ambiguous information into building new mental frameworks (mappers) versus those who take this information and attempt to categorize it into older frameworks (packers). Whether this reflects personality types or simply ways we tend to think about certain problems when it only comes to our benefit, this mapper versus packer orientation is something we see every day.
Programmers seem to think that the mapper/packer orientation is something you learn in school. As you grow up, teachers tend to take our natural development, in what is originally an entirely mapper-like way of acquainting oneself with the world, and turn it into packing by making education the process of writ memorization. According to the programmers, one can reacquire the mapper orientation with enough training. In that perspective, these orientations could be something of a plastic personality type, something learned early on and difficult to drop later. It could be representative of something picked up by those who are neuroatypical (hence programmers all seem to think they are mappers). I simply think like most cognitive concepts, these mindsets could also simply be components of an ideology or simply something we all conditionally employ depending on whether it benefits us to do so. It’s worth noting the original blogpost states, “We are all mappers, no matter how little we use the faculty.”2
In any event, this is not important. What is important is that the packer/mapper distinction has serious implications for the ways in which people collectively interpret things. If this viewpoint is true, that mappers integrate new information and that packers categorize it, then we have a pretty serious issue for the way things are not only described and understood.
Packers and Mappers: An Illustrative Proof
To elaborate how mapper/packer orientations can lead to different ways of viewing the same phenomenon, I’ll give a visual example here (Fig 1). Let’s assume there is a world where shapes categories are solely constructed by blue squares and orange rectangles. Pretend colors are fundamental to categories and you never see orange squares or blue rectangles. How would mappers and packers describe this?
A mapper will say: there are two shapes, conforming to the rule that there are blue squares and orange rectangles. There are no orange squares and no blue rectangles. Rectangles are distinguished from the square category on the basis that rectangles are orange.
A packer would likewise say: there are two shapes, conforming to the rule that there are blue squares and orange rectangles. There are no orange squares and no blue rectangles. Rectangles are distinguished from the square category on the basis that rectangles are orange.
It seems the two groups agree on the same thing, which is that one of the fundamental ways you can distinguish these two shapes from another is that one is orange and the other is blue. These appear to be inherent properties of the shapes, by definition.
Imagine one day a square turns orange, and reality needs to be reassessed (Fig 2). This is where the mentalities divide.
In reassessing what he sees, a mapper is more apt to update the parameters of the categories (Fig 3). There are now several valid shape categories, maybe there are now even six. Sorting and use of these these depends on the sorting context, but there are now at least six valid categories to be employed.
Where is our packer? The packer is still at the original framework (Fig 4). The presence of the orange square-like ambiguity has been noted, but it does not match the definition of a square or rectangle. It cannot be “packed” into either category, but is also not a separate component of this object space.
I have spoken a few times about the divide between processual and non-processual approaches to the sciences. The term processualist is a creation of post-1950s archaeology which arose out of a desire to not just study the movements of people, but the ways in which they lived. Prior to the advent of processualism, archaeologists were in the role of simply categorizing which types of cultural artifacts belonged to which group and where these groups moved before history. In an effort to free the field, several revolutionaries decided to ground archaeology in anthropological theory and to study what people lived like and how sites were formed. That is to say, they moved from looking at categories and moved to processes.
I strongly believe that this processualist/non-processualist dichotomy can be applied to a wide range of scientists and the ways scientists think about things. My friend Razib Khan, who unfortunately deletes his tweets, once commented on the divide between geneticists like Jerry Coyne, who says there is such a thing as biological human races3, versus anthropologists who say there is not one by highlighting the processualist/post-processualist divide: when anthropologists tell Jerry Coyne that there is no such thing as an intrinsic human races because you can split humans into 5 discrete groups or into 100 discrete groups, the scariest thing for Jerry Coyne to say to them is, “I agree, but all of these groups are real.”
How is this sort of an agreement even possible? In a way, for people like Coyne, these things are conceptual scaffolds. As Razib puts it, the geneticists more apt to use these categories are people who study genetic processes and do not care about categorization. They are using the term as a point of convenience, not as a statement regarding important facts about the world. This is because processualists see the important facts about the world in the ways that different components of it interact, not in what those different components are. In an ironic sense, it is the anthropologists fighting Coyne who care a great deal more about categorizing humans than he does!
This form of categorical denialism in which anthropologists will argue that race categories aren’t real simply because there are too many potential categories is textbook packer mentality, but it’s not the only place we see such a mindset applied to humans.
Over the last several months I have probably spent miles worth of finger twitches arguing over the concept of sex categories. The main question is: are there only two biological sexes? I’m not going to rehash the arguments made in my Areo piece on the topic, but here’s what’s important is that categorization here depends on certain factors. In an argument with biologist Colin Wright, who maintains there are only two sexes, I stated this:
Many peoples didn’t know anything about anisogamy and nonetheless categorized humans by biological categories: penis, vagina, and other. What would you call this?
I call this genital sex. It’s a real-world grounded category in of itself that’s not anisogamic sex but is most definitely not gender, either.Letter Conversation 113: On Adaptationism, Natural Sex & More
Despite my argumentation, I haven’t been able to convince people in Colin’s camp that a sex category based on what genitals you have is inherently a biological sex category, just as a sex category based on what type of gametes you produce is also a biological sex category. Both are very real biological categories, but which one you choose depends on which criteria are in operation. For some reason, calling both categories the same thing seems to trigger a jingle fallacy (this is where people assume two different things are the same thing because they have the same name) for some people. I think is a specific instance of packer-like thinking.
Philosopher of science Bill Wimsatt, whom I also borrowed the term scaffolding from, defines complex objects as, “systems with multiple partially overlapping boundaries with richer possibilities for interpenestrating interactions.” Which is to say that the more separate but valid ways you can look at something, the more complex it is.4
I think it’s safe to say that a lot of things are complex, and this is precisely what makes categorization so difficult. How is it that something can be one thing and something else? In what sense can something come to be referred to as too many things and therefore end up being nothing, like some people say historical and scientific revisionism is doing to us? I’m not really sure of the answers to those questions, and they probably belong to philosophers, hence science needs philosophy. But what I can say is that the ways in which we tend to think about complexity are certainly bounded by at least a few cognitive parameters, including some of which likely reflect the difference between what programmers are calling mapping and packing.
Like my blog and want to support my writing? Please consider supporting me on Patreon.
1 Carter, Alan G, and Colston Sanger. “Thinking about Thinking.” The Programmer’s Stone. DataPacRat, October 20, 1997. https://www.datapacrat.com/Opinion/Reciprocality/r0/Day1.html.
2 “Mapper (Idea).” Everything2, January 8, 2001. https://everything2.com/user/dmd/writeups/mapper.
3 Coyne, Jerry. “Are There Human Races?” Why Evolution Is True, February 28, 2012. https://whyevolutionistrue.wordpress.com/2012/02/28/are-there-human-races/.
4 Wimsatt, William C., and William Kurtz Wimsatt. Re-engineering philosophy for limited beings: Piecewise approximations to reality. Harvard University Press, 2007, 354.
One thought on “Packers and Mappers”
Wow, I didn’t have the patience to more than skim the second half… I did skim it without stopping to understand…and only had patience to understand some points in first half out of frustration and need to come post down here.
My instinct to categorize would be to look for patterns then define a specific and an other. I always anticipate future iterations on any problem… what if I have a third kid or add pets to family members
A square has two equal sides. We have blue squares. So something that has two equal sides AND is blue is a BS and to start with, the others are Others.. unless we have data otherwise others default to two positive dimensions parallel lines and orange… but we’d need a field for number of sides and im forgetting my shape equations 40 years out from using them.
To start shapes are the object and they need sides. An array of side measurements one could take the length of to get side amount and an array of angles belonging to shape object wirh shape.angle being relation between shape.sides and shape.sides
Shape.color originally default to orange but could be tested against an array or rgb format regex later. The shape object could be expanded to have music and texture and textfield and propername
GameData might have fields
ShapesGiven would be set variable array
It would be easy to count # of blue squares # of other, # of squares not blue number of orange etc, and up to programmer at anytime to answer blue squares #orange other #orange rectangles etc but could also cluge all into other without breaking model
GameState might store counts if you didn’t want look at every piece over and over for display
People? Id make women the specific case and men the catch all .. a clito easier classified a type of penis and ovaries map to testies than a womb,uterus etc.
So…am I a mapper or a packer?