Island closed on account of bears.
That was the message I got earlier this summer, when my husband and I set off for our annual trip to Lake Superior's Apostle Islands. This national park is lousy with black bears. One of the islands, Stockton, is the bear equivalent of Manhattan — boasting one of highest black bear population densities in North America. Ironically, though, that wasn't the island the bears had shut down. Instead, park visitors had been temporarily banned from setting foot on Sand Island, 4.5 square miles of forest and sandstone cliffs, shaped vaguely like a fat, upside-down comma.
The problem at Sand Island wasn't an abundance of bears. In fact, the issues could be traced back to a single bear, or possibly two, Julie Van Stappen, the park's chief of planning and resource management told me later. But rangers were worried about those bears' behavior. Within the span of few weeks, human food had been scavenged, human tents damaged, and on one memorable occasion a bear tried to snuffle its way into a tent with the people still inside. Somewhere along the line, the bears of Sand Island had been conditioned to seek out humans, rather than avoid them.
This behavior puts the humans in danger. Big as they are, bears are easily startled by humans, and a startled bear is likely to attack anything it perceives as a threat. Six people in five different states ended up on the business end of a bear mauling just this month.
Rangers hoped that, by closing the island, they'd get an opportunity to recondition the bears — effectively teaching the animals to behave like proper, human-shunning wild creatures once again.
But first, they had to invite the truant bears to a barbecue.
All you need to condition an animal is a cue and a reinforcer — let them hear or see or smell something strange and link that something with a positive experience. The most famous example, of course, are the dogs trained by 19th-century physiologist Ivan Pavlov to associate the cue of a bell with the reinforcement of food. After a while, the bell would make the dogs drool, whether or not the food ever actually appeared.
That's classical conditioning and it's more about creating a reflex rather than getting the animal to do anything active. But, frankly, food is a great motivator for all kinds of behavior, including active learning. And that's especially true for animals with varied diets like dogs … or bears. When a bear learns to associate the presence of a colorful plastic tub with the discovery of tasty food, that's really just a minor variant on classical conditioning, says Ralph Miller, a psychologist at Binghamton University. Like Pavlov's dogs, the bear is learning to associate food with a cue. If it keeps finding food in human objects, that connection between humans and food gets reinforced.
In the places where nature and human civilization butt up against one another, that fact translates into some big risks. Bears have three key activities in life. They breed, they hibernate, and they eat — and the first two activities are very dependant on how successful they are at the third. Because of that, bears are resourceful, willing to try new things, and willing to take a few risks if they get a reward for it, said Carl Lackey, a biologist with the Nevada Department of Wildlife. When bears cross paths with humans and aren't chased away by bear spray or whistles, they quickly learn that humans aren't something to fear. When they find food in those humans' trash cans or tents, the bears learn to associate people and people things with the presence of a reliable meal. Those lessons increase contact between bears and people, which, in turn increases the chances of what rangers call "bear-human conflicts". That's dangerous for people, obviously, but it's also dangerous for bears, who could end up shot when all they really wanted was your leftover hamburger.
But how do you get a bear — an animal that really wants food — to avoid an easily accessible source of food? One option is aversive conditioning, a system that tries scare bears away from people by using the same psychology that brought them to the trash can to begin with.
The Sand Island barbeque was a trap. Rangers filled a grill with bratwurst and made sure the tantalizing scent of grilled meat wafted beyond the campground area and into the woods. But they never really intended to share. Instead, they were trying to lure bears into human areas so that they could then chase the bears away from those areas using harmless little firecrackers called "whizz-bangs". The hope: To make the bears associate human beings with something scary, instead of something enjoyable.
Conditioning can involve positive reinforcers (things that attract an animal) or negative reinforcers (things the animal will avoid). When the animal gets a positive reinforcer as a reward for their behavior that's called positive reinforcement. When a negative reinforcer is removed as an award for the behavior, that's negative reinforcement. Aversive conditioning is based on the idea that, if the positive reinforcer of food is creating a cycle of positive reinforcement that brings bears closer to people, maybe you can flip the situation around. Negative reinforcement — training the bears to learn that, if they avoid humans, then bad things won't happen to them — could override the positive reinforcement of food.
There are a lot of different ways people go about this and some methods are more science-based than others. A lot of times, it simply involves waiting for a bear to show up in a place where it's not supposed to be. Then, you scare it away, using whizz-bangs, or rubber bullets, or Karelian Bear Dogs, a breed that can be trained to chase bears from human sites and to give up the chase before they injure the bear. Miller says that's not really even a reinforcement as much as just punishing animals.
In the 1980s, rangers in Yellowstone National Park tried a more Pavlovian approach. They captured known nuisance bears and put radio collars on them, so they could be tracked. When the bears came looking for food, the rangers would play a bird call — from a species that didn't live in Yellowstone — and then shoot the bears with a rubber bullet. Over time, they thought the bears would start to respond to the bird call alone, immediately leaving any area where they heard it.
But it didn't work. Not only did the bears never really develop a reliable conditioned response to the bird calls, they also didn't change their behavior in response to the rubber bullets — at least, not in the long term. The rangers found that they could scare bears away from a site, and maybe make the lesson stick for a few weeks or a season. But, eventually, the bears would start coming back.
That matches up with what Ralph Miller told me about how reinforcement works. Creating a fear stimulus that makes a bear flee when confronted with bird calls is great and all, but it's also a temporary thing. That bird call won't be there every time they encounter humans. It won't be there every single time the bear finds human food. Meanwhile, the positive reinforcement cycle that tells bears that people = food is much more reliable and constant. If you actually want to break that cycle, you have to create a new reinforcement cycle that's just as permanent as what it's trying to replace. After all, depending on the specific bear, you could be competing with a long history of training that began at birth. Carl Lackey told me that bear mothers who frequent human areas train their cubs to do the same.
There's a lesson in the story of Sand Island — taking away the positive reinforcer of human food and breaking the positive reinforcement cycle is more effective than trying to create a negative reinforcement cycle that involves sporadically frightening bears away from that food. That's actually why national parks have developed such stringent rules about what visitors are supposed to do with their trash, how they're supposed to store food, and how they're supposed to behave in bear country. It's all about removing the positive reinforcer.
In fact, the most successful reduction in bear-human conflicts is based on his principle. In the first half of the 20th century, Yellowstone National Park averaged 138 bear-caused property damage incidents and 48 bear maulings every year. It was also a time when bears were getting LOTS of positive reinforcement from human behavior. People hand-fed bears along the highways. They left food in coolers and picnic baskets where bears could easily reach it. There were even bleachers installed at local trash dumps, so people could come and watch bears eat their garbage. Generations of bears had learned — go where the people are and you will be fed.
Then, beginning in the late 1960s, Yellowstone started to change. Dumps were closed. Trash cans were bear-proofed. Rangers educated campers about securing their food and started cracking down on tourists who fed the bears. There were still problems with some of the older bears, but the younger ones learned — there's no reward to be had for pestering people. Today, despite the fact that human visitors have tripled, Yellowstone's bears are involved in fewer than a dozen incidents of property damage every year and average only a single mauling. In the end, the best way to retrain bears is to retrain people.
Edit: This story has been changed from its original version. I got some of the specifics wrong in explaining positive reinforcement and negative reinforcement and have updated the piece to reflect a more accurate explanation.
