Grazing data. Easy to see but hard to get.

Grazing by horses is a good example of the challenge of getting good data to do research on. Getting good data is often the difficult and expensive part of doing research and grazing is no exception.

As we have discussed, our small observation herd of horses grazes in a way that is both active and mysterious. They are seldom in the same location for very long (but sometimes they are). The research question is "what rules explain this complex grazing behavior?"

In order to answer this question (if we can) we will need good data on where the horses graze over time, because that is what we are trying to explain. We probably need at least several months of data. This data sounds simple enough but getting it will not be easy. It is one thing to see that the horses are typically moving around a lot, but quite another to track just where they go and when.

Here are some possibilities, ranked by the quality (and cost) of the resulting data.

1. Attach a GPS transmitter to each horse. Each transmitter sends the horses location to a central receiving station, either in real time or periodically.

2. Attach a GPS recorder to each horse, which is then read manually from time to time to get the data.

3. Contract for satellite photos of the fields to be taken at (hopefully regular) intervals. Does not work at night or when cloudy.

4. Set up automated cameras at various locations to take repeated pictures of the fields. The location of the horses at each time can then be calculated manually using triangulation. Laborious and does not work at night.

5. Monitor one or more specific locations, such as with an automated camera taking periodic pictures. Incomplete data.

6. Manually estimate and note where the horses are, from time to time. Laborious and incomplete.

There are other data collection strategies, I am sure, and I would be happy to hear about them.

In short there is no easy and cheap way to get good data, even though the behavior is right there in front of us. This challenge is a big part of what makes good science difficult and expensive to do.

Swarming horses

I have written before about the fascinating and mysterious behavior that horses exhibit when grazing together. The individual horses are frequently close together, but not always. They clearly choose collectively where to graze, which is highly variable and unpredictable, but how?

The relatively new artificial intelligence field called "swarm intelligence" may be useful here.
See http://en.m.wikipedia.org/wiki/Swarm_intelligence for an introduction.

Swarming in this case is a technical term. It refers to complex collective behavior governed by relatively simple rules operating at the individual level. Work in this area goes by other names as well, such as flocking. See http://en.m.wikipedia.org/wiki/Flocking_(behavior).

The research question then arises whether these two collective grazing behaviors of horses, staying together and moving around, can be explained using swarm or flocking rules?

For example, it has been suggested that the way deer move around from day to day is deliberately random, in order to deter predation based on predictable behavior. Where horses graze, from hour to hour and day to day, is certainly complex, but is it random or due to some combination of simple rules?

Mind you we may not have to actually build swarming horse robots. Computer simulation may be enough. We will of course also need specific data describing these behaviors in detail. Thus the research is not simple. But the results might be useful in both horse and land management, as well as for grazing critters besides horses, such as cattle, sheep and deer.

Robotic nest building

Technically what we are doing here is called cognitive ethology. Ethology involves looking for patterns in animal behavior, especially natural behavior as opposed to artificial laboratory experiments. It often involves looking at various different species as well.

The cognitive aspect is that we are looking specifically at behavior that involves understanding, reasoning and decision making. We have been studying these features in humans for many years.

Part of our approach is to look at artificial intelligence for help, because this field often involves the mimicking of cognition. As pointed out previously, building expert systems is a way to look at the rules that people and animals use to make decisions. The same is true of robotics.

Building robots to perform complex tasks has also taught us a great deal. So, for example, what would a robin's nest building robot look like? In particular, what cognitive functions would it have to implement? At the highest level these include finding and choosing a nest site, then finding and selecting various building materials, in parallel with assembling them into the final nest.

None of these are simple tasks and more importantly none is well understood. Mind you I do not expect anyone to spend the millions of dollars it would take to build such a robot. The point is to think about all the things it would have to do.

In recent years people have developed robotic cars that work pretty well under certain conditions. One wonders if a nest building robot would be simpler, or more complex, than a self driving car? I do not think we know at this point.

Eating versus investigating, two clear cases of decision making

It is important to recognize situations where animals clearly have a decision to make. These constitutes experiments of sorts, that is controlled conditions where behavior can be clearly observed. It does not matter that these conditions are created for other purposes.

Recently we have created two situations in which our observation herd of horses had to choose between eating and what I will call investigating, for want of a better term. In both cases they chose investigating, which is interesting.

In the first case we let the herd into a small field for the first time. It is fall and the grasses in their regular fields has been eaten down so that it is very short, just an inch or two tall inmost places. In fact we have begun feeding hay when the weather is bad.

The new field was thick with grass, over a foot tall in most places. The horses have never been in this field but it is across a fence from one of their regular fields so well known to them after several years of seeing into it. This field is also rather complex as it has two cross fences with gates in different places.

When we turned the horses into this new field they clearly had two options. They could start eating the new grass or they could explore the field. They chose to explore the field, in a very decisive manner.

They stayed together and covered the territory, getting into every section and corner, which took about five minutes. They did not eat during this time. Then they suddenly stopped exploring and started eating. We have now put them into this field four times and the investigating behavior has not been repeated. They know where they are it seems.

As an aside, they have done a lot of what we call grazing in motion, so the grass is being clipped pretty uniformly in the entire field.

The second case arises when we feed hay. There are seven horses in the observation herd, so we scatter the hay in at least that many small piles. The horses could simply pick a pile and start eating, perhaps after sorting for dominance, but that is not what happens.

Instead, the horses actively investigate the hay. It appears that every horse looks at every pile but it would take careful video analysis to determine this. There is certainly a great deal of milling about among the piles. I have seen this behavior repeatedly and will study it more closely.

These two cases illustrate a simple point. If we recognize when animals have big decisions to make we can study their decision making.  It can be surprising.

Migration decision mystery

The blue jays are migrating past our observatory these days. They come by in groups of from five to forty and have been doing so for several weeks; perhaps a thousand have passed by so far.

There is a great decision making mystery with the blue jay migration. It is well known and described here:
http://www.allaboutbirds.org/guide/Blue_Jay/lifehistory
And here:
http://en.m.wikipedia.org/wiki/Blue_jay

It appears that individual birds decide each year whether to migrate southward or not. There are large numbers of migrants every year but any given blue jay may or may not be among them.

The question is how each bird makes this annual migration decision? For example, what factors does the bird consider?

It has been suggested that the weather and the food supply may be important. This may be true but it does not explain why some birds in the same place leave while others stay. It may have to do with each bird's personality, which might be quite complex. Perhaps the birds that migrate are literally those who feel like doing so.

Another possibility is that these are actually group decisions. Birds make a lot of collective decisions, in ways that are poorly understood. How a flock decides which way to fly, without a leader, is the classic example of group decision making and it is poorly understood. Blue jays migrate in groups. Perhaps they also decide who will go that way. (Horses also make group decisions, such as where to graze.)

The point is that blue jay migration is a clear case of something that is actually widespread and poorly understood, namely animal decision making.

Funding my research on horse and critter cognition

Our research concept is this: Instinct is a body of expert knowledge or belief. The research question is how can we discover and represent it? After all, knowledge representation in humans is a well developed field.

So we have begun to explore the ways that this research might be funded. There are a variety of possible funding sources, because animal behavior is also a well developed field. We have done both federally funded and privately funded research in the past and both of these sources are available for animal behavior.

On the US federal side both NIH and NSF have programs related to animal behavior. The focus at NIH is veterinary, which includes instinct  driven behavior that is harmful to the animal. At NSF the topic is part of the Biological Integration program. The USDA may do some work in this area as well, but it seems to be confined to agricultural animals.

Private sector research funding comes from a variety of sources. There are foundations that fund animal research, including several for horses. There are also various membership clubs and associations, especially for domestic and sporting animals, including horses.

There are also a number of associations that focus on wildlife and conservation, both of which involve behavior. This sort of research is also done by parks, zoos, conservation preserves, etc. There may be a U.S. Federal research role here as well, because there are a number of conservation and land management agencies.

Another interesting possibility is the emerging practice of "crowdsourcing" or "crowdfunding" where large numbers of people each contribute a small amount to a project. In particular there are a lot of horse owners who are interested in understanding their critters.

Animals lead full lives that we do not understand

Lately we have been watching a pair of Carolina wrens feeding their nestlings. There is nothing unusual in their behavior, with one exception. This is that they spend a lot of time "singing" or as I prefer to call it, in calling.

Neither term seems correct because each suggests aspects which may not be true, because humans sing and call for specific reasons. Perhaps vocalizing is the best term for it.

In any case this behavior seems odd. Feeding young is a time consuming and strenuous activity. Why do the birds also spend a lot of time vocalizing, which is also time consuming and strenuous? Why not just feed the young, then eat and rest?

The point is that feeding young has a clear purpose, while vocalizing during feeding does not. Understanding the purpose of behavior is central to understanding that behavior and this is the grand challenge when it comes to understanding animal behavior. Recording the behavior is not enough, although it may help reveal the underlying purpose.

Note that explanations of human behavior are typically in terms of purpose, not action. "Going to work" is a better explanation than "driving a car" when someone asks what you are doing. The why of behavior is often more important than the what of it.

Every day I see behavior in animals that I do not understand because I do not know its purpose. It is important to learn to see things this way, to recognize how little we understand and to figure out how we can come to understand it. Moreover, simply saying that the behavior is instinctive is not an explanation at all.

Instinct does not control behavior

Saying that an animal's behavior is instinctive is deeply ambiguous. It incorrectly suggests that the instinct is in control, but while knowledge often guides or assists behavior it does not control it.

Instinct is a form of knowledge (or belief) and of course what we know or believe has a lot to do with what we do in a given situation. But this does not mean that we do not think or make decisions. Yet the concept of instinct is often used to imply that the critter does not think. This is a deep conceptual confusion.

In fact instinct allows the animal to think more deeply that it otherwise could, because that is how knowledge helps us. Here is a simple example. I have been treating a horse for a nasty wound, following the vet's instructions and working on the wound twice a day.  Recently there was a new swelling, which concerned me, so I called the vet to come take a look. He looked at the wound and quickly saw what I could not see, which is that it was healing nicely. The swelling was good, not bad, a normal part of the progression.

Because of his knowledge the vet literally saw what I could not see. This is how knowledge works, and how instinct works as well. Knowledge lets us see (or smell, feel, etc.) the world in ways that those who lack the knowledge cannot.

So, for example, a bird can see a good nesting site, or some good nest building material. A horse can see a good thing to eat or a good place to scratch its back. A beaver can see that it now needs some mud. In no case does this mean that the animal is not thinking, quite the contrary. It is actually able to think in ways that we cannot, because it has knowledge that we lack.

Of course the animal can also be wrong, just as humans often are. This is why I sometimes mention belief along with knowledge, because the animal may be working on a false belief. For example, horses are wary animals by instinct and they may shy because they see what they mistakenly think is a threat, such as a stump. Even experts make mistakes.

In horses the idea of misunderstanding or false beliefs on the horse's part may be important when it comes to training and managements problems. But first we have to understand the underlying instinct.

Modeling Instinct Using Expert Systems

Given that instincts are bodies of expert knowledge then the methods of expert system building should be useful in modeling them. In fact trying to build an expert system model of an instinct might be the ideal way to come to grips with the deep research problem of how instincts work. Modeling instincts is a research program.

A bird building a nest or a beaver building a dam are clearly cases of the application of expert knowledge. If anyone doubts this they should try building a bird nest. So are a horse grazing in motion, a crow cawing, a herbivore selecting plants to eat, or a woodpecker picking a drum tree, as I see it. Each of these cases is discussed in prior articles here. There are many other cases that could be mentioned. In fact defining the specific areas of instinctive expertise in each kind of critter is a research program all on its own.

Building an expert system primarily means specifying a set of rules that embody a specific body of expert knowledge. That this can be done, that such expert rules exist, is itself a great discovery. Often the individual rules are relatively simple. It is the combining of these simple rules that yields the complex behavior of expertise.

Normally the expert rules are found using a process called knowledge engineering. It involves a combination of interviewing experts and reading technical documents, such as manuals, handbooks and textbooks.

When it comes to horses or critters in general, this method of knowledge engineering is not available. Rather the approach has to be to ask what expert rules explain the observed behavior. This is likely to be significantly more difficult than simply interviewing an expert, but there is no reason it cannot work. Even asking the question is useful because it creates a systematic approach to understanding an instinct. It breaks the problem down into one of finding a set of simple rules.

For example, here are two simple rules that might help explain how herbivores decide which plants to eat:

Rule 1: If it tastes good it is probably okay to eat it.

Rule 2: If it tastes bad it is probably not okay to eat it.

Note that the use of "it" here will require being able to tell one kind of plant from another. This will require a set of rules of its own. The use of "probably" means that other rules might modify these rules, such that good tasting plants are not eaten and bad tasting plants are eaten.

The point is that instinct is often an innate form of expertise. Trying to find the simple expert rules underlying important instincts is a feasible research program. This will greatly increase our understanding of animal behavior. In the case of horses this understanding will facilitate the training and management of the critters.

What plant eaters eat and how to think about it

We recently observed some profoundly interesting eating behavior in groundhogs. A juvenile groundhog and its mother were grazing together. The mother started eating a plant and the juvenile came over and ate some of it out of her mouth. It then went and ate a great deal of the same plant, which we had not seen it do before.

It is possible that this was all a coincidence, but it looks like the juvenile learned that this kind of plant was edible from its mother. This raises the issue of how herbivores, including horses, know what to eat? It also raises the issue of how animals know how to learn?

The eating issue is interesting because the number of different species of plants is enormous, so instinct alone cannot say which are edible and which are not. Instinct can provide general guidance, beyond which there must be some some sort of learning process. but the learning process itself must be at least partially instinctive.

Here I am reminded of Chomsky's theory of language learning in humans. He argues that infant humans learn language far to quickly for the process to be one of inductive inference, that is by generalizing broad rules from narrow instances. There are too many different possible languages that fit the infant's limited experiences.

This also rules out trial and error learning, which we do not observe. Trial and error may occur for specific words, but not for learning the language as a whole.

Chomsky therefore concludes that all human language has an underlying structure that is known instinctively. If so then the vast array of different human languages are merely local variants on this universal underlying structure. Thus the child is not learning language per se, rather just the local variant.

The same may also be true when horses, groundhogs and rabbits learn what to eat. The basic framework knowledge must be instinctive, supplemented by a learning process that is also grounded in instinct. This is just two examples of an instinct being a body of basic knowledge. The challenges are (1) how to figure out what that knowledge is and (2) how to express it using human concepts and language, which may be very different from the critter's concepts.

Selecting a drum tree

I have long been fascinated by woodpeckers drumming on trees because the tree is basically an instrument. This is a good case of (1) an animal implementing an instinct and (2) the thinking required on the animal's part.

The big thing is selecting the tree. Drumming is generally considered to be the equivalent of bird song, so it proclaims a territory. So the first thing the woodpecker has to do is to pick a territory. My guess is that this is based on picking a nesting site. Given the nest site the choice of a drum tree is constrained by distance. The woodpecker cannot just go off and find a good drum tree somewhere.

A good drum tree is probably relatively rare. It is typically a standing, but still hard, dead tree with no top and little or no bark. Having rotten wood or a top or bark would all dampen the sound of the drumming. I doubt that the woodpecker tries every tree so they probably know what they are looking for by instinct, but picking the right tree is a local decision.

Thus the woodpecker is solving a two variable local optimization problem, which is not simple. That is, it has to find a good drum tree that is close enough to the nest site to do the job of territorial announcement. I imagine this involves a lot of looking and testing.

If the bird uses more than one drum tree then there is also the issue of deciding which one to use at any given time. This is a separate local decision process.

It is also interesting that woodpeckers sometimes use buildings for drumming. This means that the instinct is not confined to looking for a tree. They are looking for a way to make a certain sound, which is a relatively abstract desire.

The point is that, just as with our other cases, instinct may give the woodpecker knowledge and a desire to act but local decisions are still required. Making these decisions requires thought, as well as the use of concepts.

Instinct is flexible so its execution requires thought

One of the basic points I am trying to get across is that the exercise of an instinct requires a great deal of thinking on the animal's part. Instinct is basically an alternative to learning, which means the critter knows something, or how to do something, without first learning it. The animal is born an expert, as it were.

But actually implementing that knowledge in a specific situation requires a lot of thought, just as it does with learned expertise in humans. So simply saying that an animal does what it does by instinct is missing an important question, namely what is the animal thinking? Interestingly there is a lot of research on how humans apply expertise, which may be useful in understanding how animals apply instincts.

Nor is understanding a specific instinct easy. Here the key question is what does the animal have to know (or believe) in order to do what it does? Merely saying it is instinctive is not helpful.

Beavers build many dams, but why?

In an earlier post I pointed out the complex structure of beaver dams and the many decisions that the beavers have to make in building them. there is another complex aspect to beaver dams and a scientific mystery as well. Beavers build many dams, but why?

Most people know that the beaver builds a dam, which creates a pond, and in that pond they then build a sturdy lodge with an underwater entrance. This lodge becomes the beavers home, usually occupied by a mating pair, where they raise a family.

What is less well known is that the beaver pair also build other dams nearby, sometimes a lot of other dams. Why the do this is not well understood, but I have a conjecture about the reason. More on that in a bit.

Two examples will help. In one case we saw the beavers had built five dams in a row on a steep mountain stream. These dams were massive, each almost six feet high. In another case a pair of beavers started building dams on two small side streams that were close together. They built 36 dams in all, mostly relatively small one's.

In both cases the beavers also built their customary lodge in one pond, but why go to the extraordinary extra effort of building all those other dams? Not only does this take effort but it is dangerous.

The beaver has to leave the safety of the pond to get the wood to build a dam and during that time it is exposed to predation. The more dams it builds the further it has to go, and the more time it spends exposed, greatly increasing the danger. Why do this I wondered?

When I looked into it I found that this behavior was well know but the only explanation offered was that the beaver could not stop building dams. In other words all this dangerous activity was an irrational compulsion.

To me this explanation was unacceptable because evolution could easily modify this behavior. There had to be a good reason for building all these dams. The basic principle is animals do what they do for good reasons.

My conjecture is that these extra dams clear land in what would otherwise often be dense forest. Moreover the first trees to grow after the cleared land dries are often aspen and alder, which are the beaver's standard food sources. Thus if I am correct then the beavers are tree farming, not for themselves but for future generations. I would be interested other possible explanations.

The basic point is that animal behavior may be complex and subtle, and hard to explain, but it is unlikely to be irrational on a large scale.

Our Observatory

Our home and offices are set up to watch the horses so we call it the observatory. It is situated on a rise overlooking one of the pastures and with a clear view of the other. We spend several hours a day specifically watching the horses and keep tabs on them the rest of the time, watching closely if something interesting happens.

There are seven horses in the herd, a combination of tennessee walkers and rocky mountain horses. They range in age from five to thirty and all but one are geldings. We do not ride them, in fact we handle them as little as possible, letting them be horses in their own way. There are interesting terrain and vegetation differences in the twenty or so acres they live in.

Of course anyone can watch horses and many people do. The difference is that we are implementing a specific theory of animal behavior, so we are looking for things that others might not see. Our theory and the things we see are discussed in the other posts here in the blog.

The central questions we pursue are what are they doing, what do they need to believe and/or think in order to do it, and why are they doing it? We also spend a lot of time considering why it is so hard to answer these simple questions.

By coincidence there are also a lot of mostly wild critters that we also watch. We are surrounded by a large US National Forest.

Nonverbal thinking is what animals do

Humans do a lot of verbal thinking and this makes it hard to understand nonverbal thinking. Verbal thinking means having words and sentences occur to you. This happens when we talk or write of course, but it also frequently happens without external expression, inside our heads as it were.

The concept of thinking in humans is sometimes limited to verbal thinking, in which case animals do not think. But humans do lots of nonverbal thinking as well and this is what animals do.

Nonverbal thinking occurs whenever there is intelligent behavior. An example of nonverbal thinking in humans is eating a meal, piece by piece and sip by sip, all the while talking about something else. Making the many decision required to eat the meal requires a lot of nonverbal thought. As discussed in the last post, horses also exhibit elaborate intelligent behavior when grazing. This too is nonverbal thinking.

If you watch yourself carefully you will see that you do a great deal of nonverbal thinking as you move about and do stuff. Every action involves multiple decisions.

Moreover you need a great deal of understanding of the world around you in order to do what you do, even to do the simplest things. And so it is with horses and other critters. The challenge is to figure out what the horse or other animal understands, that is, what concepts do they use when they think non-verbally?

For example simply putting on your socks requires knowing what socks are, knowing that you need to put them on, knowing where the are, knowing how to get there, knowing how to select them, knowing how to get them out, knowing where to go to put them on, and knowing how to put them on. That is a lot of knowing and a lot of thinking.

Many things that horses and other animals do are just as complicated as putting on your socks. These actions require a lot of nonverbal thought. The challenge is to figure out what this thinking is and how to describe it, especially the concepts that the critter must have in order to do what they do.

Horses make a lot of decisions

In the last post I talked about how beaver are constantly making local decisions when building their dams. Instinct may give the beaver the knowledge they need to make these decisions but it does not make the decisions for them. Instinct is a way of knowing, not a way of thinking.

The same is true for horses, only the horse's decision making may not be as obvious since they are not building stuff. When they are on their own the horse's number one activity is grazing and they spend many hours doing it. Grazing involves almost continuous decision making because you have to constantly decide which specific thing to eat next.

To see this consider the following thought experiment. You are in a pasture and you pluck a bit of grass. Now you want to pluck a second bit. How many possibilities are there to choose from? Even if you do not move there are a great many, so you have to make a local decision.

People do this too of course, for example repeatedly choosing from the different things on their plate. You can watch yourself do it. What is especially important is that this decision making typically does not involve verbal thinking, so horses can do it too. A lot of thinking is nonverbal.

Moreover, as I described in an earlier post, horses often "graze in motion" which means taking a step or two every few seconds. When to step and where also requires making decisions.

Then there is the fact that the herd as a whole (or in part) often moves more or less together. How this is done is the topic of much scientific research but it clearly involves a great deal of decision making by the horses involved.

Sometimes this decision making regarding what to eat next involves food that is not close by, or even in sight. The entire herd may abruptly go a long distance, into a new field, to graze on some bushes. Or a single horse may do it, leaving the herd behind.

The point is that horses make a lot of decisions as they do what they do and decision making is a form of thinking.