Monday, December 29, 2008

Now Wait a Minute ...

It was exactly the kind of search that members of our all-volunteer team are not supposed to be doing: A missing despondent, spouse couldn’t account for all the guns in the house, and a very suburban setting to boot. It’s what we simply call a “law enforcement search,” three guesses who has the training, equipment, and authority to conduct it.

Having said that, just because we couldn’t look for the subject didn’t mean we couldn’t help at all. Our folks took places in the incident command post; among other field-search specialists, Barbara Butler, then a rookie member of our team, and I accompanied a local law officer on a house-to-house survey of the neighborhood, to find out if anybody had seen our guy.

I scored some points out of the gate, by asking the officer if he knew if anybody in the neighborhood was out of town or of any houses otherwise unoccupied. He realized right away what I was getting at, and drove us over to a house that we canvassed quickly to make sure there weren’t any open or unlocked doors or windows. Dead end, but it’s always good to be pitching ideas they like [1].

It was also a bit of an education as to how the world looks to our law-enforcement colleagues: I recall our reception at one house. The officer had commented on these folks being frequent fliers, as it were, and they were bristling with hostility the instant they saw the squad car pull up. But when he introduced us as the department’s search-and-rescue team [2], and it became clear we were looking for their missing neighbor and wanted their help, not to roust them, they brightened up and did what they could. They hadn’t seen anything, but I like to think it may have helped both family and cop down the road to have had at least one interaction that wasn’t adversarial.

But the reason I bring this story up is what happened just as we were saying goodbye to those folks: the officer’s radio crackled with the news that a state trooper had found our guy, not too far from where we stood. Barb and I piled into the back of the cruiser and got quite a ride.

As we careened down the suburban avenues, lights and sirens going, other local and state police cars filed in with us until we were something of a parade. It was fantastic, like being the cameraman for an episode of Cops.

So we pull up to the verge of a small, wooded local park, and the officers pile out of their cars, running into the woods. I don’t think that at the time I knew the full story — that the subject had come at a state trooper with a knife — but it was pretty clear that the situation was their bailiwick. Not for us untrained, unarmed volunteers.

Barb is walking after them.

My hand comes down on her shoulder: “No you are not,” I say, in what probably wasn’t the only time I’ve pulled rank on a junior team member, but may be the only time it ever worked. A few minutes later, we see the police escorting the guy to one of their cars — the trooper had had a Knock on Wood [3] moment and come through with flying colors, disarming a mentally ill subject without having to hurt him.

This is not to pick on Barb, by the way: in the rush to go help, rescuers often get themselves into situations they can’t handle. Tunnel vision kills rescuers: that’s why it frustrates me when everybody lionizes a local rescuer who leaves a widow(er) and kids behind attempting something he or she wasn’t trained to do. You’re always supposed to survey any rescue scene and make sure there isn’t a situation that could make you patient number two; you’re always supposed to stay within the limits of what you’ve been trained to do — and just because people make understandable, very human mistakes, teammates are supposed to rein you in now and again.

All the best intentions aside, somebody just has to say, “Now wait a minute ...”

Enter a paper from last year on ground-scent tracking by humans, which came to my attention again because an automated PychINFO search tagged it (possibly because it just entered that database). The leaders of the multi-institutional team that did the study were Jess Porter of UC Berkeley and Noam Sobel of the Weizmann Institute of Science in Israel.

We have to keep one big thing in mind: These guys were studying whether humans could follow a ground trail, much as a search-and-rescue trailing dog does, mainly because they wanted to study the phenomenon of ground trailing with the benefit of subjects that can tell you what they’re thinking and what strategies they’re using, and will accept all sorts of nose and head attachments that might get you bit by other subjects. On these terms they succeeded amply [4].

The idea was simple: Could they produce a trail of a scent that humans can detect — chocolate — that humans could follow, without the use of their eyes or other senses? And what would that experience teach about how other species follow ground scent?

The short answer: People could not only learn to follow a chocolate scent on the ground, they could get much better at it with practice. Even more interesting:
* Just as in dogs, a higher frequency of sniffing allowed a more accurate and speedy trail.
* As in critters as different as maggots and rats, both nostrils have to be in play for a good, efficient trail, further cementing the human subject as relevant to what the animals are doing.

The latter observation was particularly interesting from the SAR perspective, because it isn’t at all clear how two nostrils, right next to each other, can give you any useful information about which direction to move along a trail. How can a dog’s nostrils, maybe an inch apart, tell you anything about the age of scent in either direction when the microscopic skin particles that carry ground-based scent are only a fraction of a second older than those an inch farther down the trail?

The human subjects simply told Porter, Sobel et al. how they did it: They worked along the edges to keep themselves oriented, in much the same way that a smoke-blinded firefighter sticks to the wall to keep form losing his way in a burning building. And much the same way a number of species follow airborne scent plumes. And for that, the small separation of vertebrate nostrils is more than wide enough, as the outside edges of a scent plume can be fairly sharp.

So this pretty much looks like a win for nurture over nature, right? Maybe humans can become good enough at scent tracking to do without the bloodhounds? Even the researchers said their findings “... suggest that the poor reputation of human olfaction may reflect, in part, behavioral demands rather than ultimate abilities.”

The moderation of that statement with “in part” aside, there’s only one problem with this idea, and it’s where the nurture-conquers-all take starts to fall apart. Instead of leaving trace scent on the ground, which is what SAR dogs follow, the researchers used a string soaked in odorant to create their “ground scent.” That means that, instead of the scent impression left behind by a moving point source of scent, which is what a SAR dog has to trail, their subjects were traveling along a linear source of relatively constant scent. The ersatz trail doesn’t vary nearly as much, if at all, in age and, presumably, intensity as you go along. This isn’t immediately important to the researchers’ findings, but it does cast a shadow on the the long-term relevance (though they could re-design the experiment to get around this).

A string soaked in chocolate is a much stronger, more consistent scent source than a moving human being, whose microscopic debris are hitting the ground in a ragged path partly dictated by crosswind and possibly varying quite a bit in intensity as it goes along. The humans may be able to do the scent-string task, and they may even get good at it: but it’s not the same job the dogs do. And when you compare what humans can do with a continuous, presumably strong scent source versus a dog following a trace trail that’s a day or two old — this experimental design can’t even touch the subject of trail aging realistically — and you realize that the dog is achieving something vastly harder.

Worse, the information this experiment gives on how the human subjects followed the scent trail may not be relevant to real-life scent trailing: because the trail doesn’t vary along its length like a real one does, the directional cue is vastly different. The researchers dont say, but it looks like their subjects basically knew which way to turn once they hit the trail, they just needed to know they’d hit it — and to stay with it once they did. As such, the situation is far closer to that of detecting airborne scent plumes, as in the research I cited above and as airscenting SAR dogs do, than it is to ground scenting. That’s because in the former, there’s also a cue as to which direction to head when you hit the scent: the direction of the wind. For modeling ground scent, it maybe isn’t so germane [5]:

[Note that the scent actually takes the form of delicate filaments, not the dots I have here — I haven’t figured out how to draw that effect yet.]

Which doesn’t really constitute a serious critique of the present experiment — as I said, I think they can redesign it to be more realistic, and in any case they succeeded in meeting their initial goal of showing humans can trail at all. But it does remind us that something is going on between the dog’s prominent nose, ample receptor repertoire, massive olfactory bulbs, and who knows what else, that transcends anything our environment can do to help us use our smellers better.

And Barb? Well, she’s been with the team for quite a few years now, and unlike your’s truly, actually went through the official training and certification for incident staff members. At a search, she’s the boss of me.

If you want to know whether she’s ever had to rein me in, you’ll have to ask her.

[1] And you can bet that, if we’d found anything important, I would have happily given him the credit — local law enforcement are our clients.

[2] No, we’re not actually affiliated with that department. But I never argue when they take ownership of us — again, they’re our clients, and it’s a good thing when they want you. Our uniforms, by the way, were designed to look decidedly not like police uniforms.

[3] The version of the song I knew was
The Mighty Mighty Bosstones’, and it’s great — but I was surprised to find that it has a 42-year history I hadn’t known about.

[4] I’m not going to get into the difference between tracking and trailing as dog handlers use the terms, partly because you’ll get different answers from different handler, and partly because the design of this experiment obscures the issue.

[5] Without going into a ton of detail, a scent plume in the air — and probably a trail on the ground, since it’s in effect an imprint of earlier airborne scent — is pretty ragged and intermittent. Though it may seem obvious that you head into the stronger scent, in practice the direction of the stronger scent may be anything but obvious.

1 comment:

Smart Dogs said...

If the chocolate researchers used premenstrual women instead of college students to conduct their study - they'd probably have performed like Bloodhounds (or better yet - English Shepherds).

Also, in previous life I did HAZMAT response work. Was the supervisor / incident commander and did annual OSHA training for responders. One of the things I somewhat fanatically drilled into people's heads was the absolute necessity of NOT RUSHING BLINDLY INTO A SITUATION. I was a major bitch about it.

But - no one ever died at my sites.