What Does Your Dog Think a Trail Is?

A while ago, I was chatting with a friend from a mantrailing background about my tracking training with Merlin. She asked me a question that I couldn't answer straight away:

“If you lay a track for your own dog, why doesn't the dog simply turn around and indicate on you after taking scent from the article?”

It's a great question. After all, you're right there. The dog knows where you are and can presumably smell you. So what makes a tracking dog set off in the direction of the handler's track instead? What got me was that I couldn't answer the question fluently or articulate the answer with enough clarity. And that was annoying! I simply hadn't thought about it. Mostly, I'd been too busy doing "the work". But thinking it through led me to a much bigger question: What does my dog actually think a scent trail is?

My experience of teaching Merlin to track after a year or so of intensity-based mantrailing made me think much more carefully about this. Not because the question only applies when changing disciplines, but because changing the task made some of the learning involved much more visible. I've come to think about it in terms of two things: context and continuity.

Context: what problem are we asking the dog to solve?

One of the interesting things about training tracking or mantrailing is what happens when we change what has value. Take articles and people. A green dog that has only ever worked article tracks learns that articles encountered along the track are significant. We deliberately build value in those articles and teach the dog what to do when they find one. If we later put a person on the track and effectively make that person the final “article”, the dog doesn't necessarily know what to do with them. From our perspective, that can seem strange. The track was laid by a person. Surely finding the person should make sense? But why would it? Nothing in that dog's training history has taught them that a person has the same significance as an article. We have changed something about the task and assumed the dog will make a connection we haven't taught.

The reverse is also true. A dog with a mantrailing background has learned that the person matters. Put an article on the trail and there is no reason for the dog to automatically recognise that this object now has value. That relationship also has to be taught. This was something I had to work through when teaching Merlin to track. His previous learning hadn't disappeared just because I had decided we were now doing tracking. At times, behaviours that had previously been successful could reappear because, from Merlin's perspective, they had a perfectly good reinforcement history.

This is where context plays a role. The dog isn't responding to scent in isolation. Equipment, start routines, scent articles, handler behaviour, verbal cues and the wider training situation can all provide information about which task is happening and which behaviours are likely to be relevant. And that matters even if your dog has only ever trained in one discipline. From the first track or trail we lay, we are creating a context in which particular behaviours, objects and outcomes acquire meaning.

Is previous scent experience always an advantage?

This also makes me question the assumption that experience in one scent discipline will necessarily benefit another. We often talk about scent skills as though they are straightforwardly transferable. A dog already knows how to find people, therefore perhaps tracking should be easier. But the dog hasn't learned to be “good at scent following”. They have learned particular solutions to a particular scent problem. Some of that learning may transfer. Some may be irrelevant. And some may compete with what we are now trying to teach.

A dog with a strong reinforcement history around reaching a person doesn't automatically understand the significance of articles. Equally, a dog that has learned to work a track in a particular way hasn't necessarily learned that there is, perhaps, a more efficient solution. This is one reason I don't generally recommend mixing and matching between mantrailing and tracking. The two tasks can share a lot of superficial features while asking the dog to learn different things within them. Previous learning can be powerful. But we shouldn't assume that it will transfer in the direction we want just because both activities involve following human scent.

A recent study by Vera Volmary made me think about this in another way. Seventy mantrailing teams were tested under randomised, double-blind field conditions, including situations in which dogs had to select the trail associated with a presented scent article from competing human trails. Across the study, initial trail selection was not reproducibly above chance [1]. That doesn't necessarily tell us that mantrailing dogs cannot discriminate individual human scent. Nor does it tell us exactly what these dogs had learned to follow. But that is precisely what I find interesting.

The dogs were trained mantrailing dogs. Many would follow trails, sometimes for considerable distances, yet as a group they did not reliably solve the specific discrimination problem the experiment presented [1]. Reading that study made me think much harder about the distinction between the task we believe we have trained and the information the dog has actually learned to use. A dog can apparently have a strong learned concept of following a trail without necessarily demonstrating, under this particular experimental set-up, the individual-scent discrimination that we might assume underlies it.

For me, that brings us straight back to context. Training creates a particular history of scent problems encountered, cues, decisions and reinforcement. Change the problem, remove some of the familiar information or ask the dog to make a distinction that its previous training has not actually required and we shouldn't necessarily expect the same performance. However, context only gives us part of the answer to my friend's original question. It can help explain why turning around and indicating on me isn't the relevant response when Merlin is presented with a tracking start. It doesn't explain how he knows which way to go. For that, we need to think about continuity.

What is scent continuity?

Research has shown that dogs can determine the direction of travel of a human track from sequential scent information. In one experimental study, dogs could determine direction from as few as five footsteps, but not three; rearranging the order of the footsteps disrupted that ability [2]. But what does that directionality actually smell like? Scientifically, we are a long way from knowing this. But I've found it useful to think about this in terms of something I call scent continuity: the pattern of scent information the dog uses to maintain progress through the task it has learned. There will, of course, be physical and chemical information underlying that continuity. Human scent is chemically complex, and as a person moves through an environment that information can be distributed in different ways. On and around the ground, around the dog's nose height and into the air, for example.

Continuity isn't intended to describe one particular chemical constituent within that scent picture. I'm interested in the next level up: how does the dog use the scent information available to it to keep solving the problem? Some of that information will be more readily available than other parts. Different dogs may access different parts of it. And importantly, the most useful information depends partly on what the dog is trying to achieve. Let me explain.

A recent study of wilderness search and rescue area dogs gives us an interesting example. Researchers separated human breath scent from scent originating from exposed skin and tested how six certified area-search dogs responded. All six dogs responded to the breath condition, while only two responded to cutaneous scent [3]. These weren't tracking dogs. Their trained task was to locate a live human source. For me, the more relevant question isn't whether this means that dogs “follow breath”, it's what the result tells us about which parts of a complex human scent picture may be particularly useful to dogs performing a particular task.

For an area-search dog, the work is about getting to the source and the source is the person. Scent information that allows the dog to maintain progress towards that live human source is therefore useful. Whereas, a tracking dog has a different problem. The work is the track itself. Articles along that track acquire value and provide opportunities for reinforcement. Therefore, following the track reliably predicts rewards and the most useful scent information isn't necessarily whatever provides the quickest route to the person. It is the information that allows the dog to maintain continuity along the track. And this is where training enters the picture again.

What continuity are we reinforcing?

One of the things I've noticed through training and instructing intensity-based mantrailing is just how differently dogs can solve what appears, to us, to be the same scent problem. Intensity-based mantrailing gives the dog considerable freedom to develop a technique that works for them, and what emerges can be quite individual. Some dogs develop very characteristic ways of working, while others alter their technique much more readily according to the problem or conditions they face. More recently, tracking has made me think more carefully about what sits underneath those differences. Even when we aren't deliberately teaching a particular scenting technique, the problems we set still shape the solutions that become successful.

Imagine repeatedly working short, fresh trails with a highly valuable person at the end. We may think of the exercise as teaching the dog to follow the trail because that is what we have laid. But from the dog's perspective, there may be several ways to solve the problem. If scent from the person is already available at distance, working directly towards that source may get them to reinforcement. If other scent information associated with the trail is useful, they may use that instead. Different dogs may discover and favour different combinations of information. That strategy isn't necessarily right or wrong. It has worked.

The same applies when we deliberately build value in a track. Articles and reinforcement along it change the problem by making maintaining connection with the track itself valuable. That doesn't mean we know exactly which chemical components or parts of the scent picture the dog is following. But we have changed which strategies reliably lead to reinforcement. So perhaps the important point isn't whether a training method explicitly tells the dog how to use its nose. The construction of the training problem itself shapes which scenting strategies are successful. This is where I think the idea of scent continuity becomes useful. The dog brings its own tendencies and the environment determines what scent information is available. The way we construct the problem changes which information is useful and reinforcement strengthens the strategies that work. Together, those things may shape what continuity means to that dog.

Continuity when things get complicated.

I became particularly conscious of this when introducing greater uncertainty and contamination into Merlin's tracking. The point wasn't to make his tracks harder, but when other scent information was introduced, the scent picture naturally becomes more complicated. The information gets weaker or more fragmented. I needed Merlin to continue moving through that complexity without abandoning the track he was following. This scenario made me think about continuity in a slightly different way. Continuity doesn't necessarily mean that the scent picture always remains the same. The dog may need to recognise what remains relevant as the picture changes. I can't know exactly what Merlin identifies as that thread of continuity. But I can influence his experience of it. The problems I set, the decisions he gets to make and where reinforcement occurs all contribute to what continues to work for him.

So why doesn't the dog just turn around?

Which brings me back to where I started. I lay Merlin's track. I present him with the scent article. I'm standing immediately behind him. But turning around and indicating on me isn't the problem his tracking context asks him to solve. Instead, his training has given value to following the continuity of the track away from that start point. I may never know exactly which parts of the scent picture Merlin uses to do that, but I can observe how he works and I can recognise that my training is helping to shape what information becomes useful and rewarding to him.

So, if we want our dogs to become more accurate or methodical, to slow down or speed up, to work longer trails, or to work through contamination and distractions, perhaps we need to look at the gap between the behaviour we see, what problem we think we have set, how the dog is responding to that problem and the behaviour or outcome we actually want. Our training may already be reinforcing a perfectly successful strategy, but not necessarily the one we thought we were training. Changing the behaviour may therefore mean changing the problem we set, where reinforcement occurs or even which information becomes valuable to the dog.

References

[1] Volmary, V. (2026). Initial trail selection in mantrailing dogs under double-blind field conditions: a statistical evaluation of scent discrimination performance. Forensic Science International: Synergy, 12, 100682.

[2] Hepper, P. G. & Wells, D. L. (2005). How many footsteps do dogs need to determine the direction of an odour trail? Chemical Senses, 30(4), 291–298.

[3] Walker, M. K., Hall, N. J. & Aviles-Rosa, E. O. (2026). Constituents of human scent as perceived by wilderness search and rescue area dogs. Applied Animal Behaviour Science, 300, 107033.