Horses cannier than we first thought

New research suggests that horses may be considerably smarter than we previously thought. The study - which was carried out by scientists at Nottingham Trent University - found that our equine friends were cognitively one step ahead of the researchers, who were initially baffled that the animals could learn a reward based task incredibly quickly, but then seemed to perform worse than a goldfish when the rules of the game changed. Until they realised that the horses had actually sussed that they didn’t need to cooperate in the next stage of the task, so they didn’t bother! Hoof’d have thought it! Louise Evans carried out the study…

Louise - I trained a group of 20 horses to touch a target, which was a piece of card. Whenever the horse touched the target with their nose, they got a treat. That motivated them to touch the target every time they saw it. To test whether they could control this impulse once they'd learnt it, I introduced a light cue. So I was wearing a cyclist's head torch and whenever the light was on, that was the horse's cue to stop, just like a stoplight on a traffic light system. When I switched the light back off again, they could touch the target and be rewarded with the food like normal.

Chris - And how quickly did they pick it up or did they not pick it up at all?

Louise - Well, that's the thing, Chris, they didn't pick it up very well at all. So we trained this group of horses for three weeks and all of our horses were doing really poorly, which was very concerning to us because we know that goldfish can do this.

Chris - For comparison purposes, how quickly did they learn the first task? As in, I present the target, if they come and touch it, I give them a reward.

Louise - They picked up that task within three training sessions. We were expecting them to be similar with this task. Actually, what we saw was that they were touching the target every single time, regardless of whether that stop light was on. We do ride horses, we sit on their backs, we put ourselves in dangerous positions with horses, so if they're not as smart as a goldfish, that is quite concerning.

Chris - Well, yeah. So what did you think was going on, then? Do you think they were just being obstinate or was there actually something about the training? They just couldn't learn this?

Louise - One of the ideas we had was, maybe horses just can't do this. They're just not able to control that urge to touch the target when there's food involved. We also thought maybe they just can't see the light very well, although we did think that was unlikely because it was quite bright. Our third possible explanation, which is what we wanted to test in this study, was whether horses had maybe figured out a clever way to cheat the game or to take part in this task in a lazy way. What I mean by that is, maybe the horses had figured out that actually, if they touch the target every single time, sometimes they get rewarded and sometimes they don't get rewarded, but nothing bad ever happens, they never get punished. Why would they bother expending the sort of mental energy it takes to pay attention to a light when they could just ignore the light and touch the target every time?

Chris - That means if you introduce some kind of bad outcome if they do it wrong, they touch the target when they shouldn't, will that motivate them more to then comply?

Louise - Well, that was exactly what we tested. We gave the horses another three training sessions, and this time when they ignored the stoplight and continued to touch the target, we gave them a ten second penalty: I stepped away from the horse and I removed the target from their reach so they couldn't access the game.

Chris - And then what happened? Did they learn it after that?

Louise - As soon as we introduced cost, the horses all started observing the light cue and following the rules and playing by the rules. What that suggested to us is that perhaps the horses had actually understood the task all along because if they were just learning from the introduction of that penalty, we would expect them to gradually improve their performance, whereas this was a very sudden switch in strategy that we observed.

Chris - It was more a case of who's playing with who by the sound of it, isn't it? They actually sussed you out. They knew what was going on, they knew the score. This elevates horses above goldfish, presumably. Does this mean you're seeing your horses in a totally new light now then?

Louise - Yeah, exactly. The really exciting part about this study is that we seem to have uncovered an ability that we didn't know horses could do. It's called model based learning. What that means is that the horse builds a mental model or a mental picture of the game, of the different actions they could take and the different outcomes that they might have. That's a really advanced skill. It's a little bit like a human playing chess. You build a mental model, a mental map of, well, if I move this piece here, then my opponent might be able to move his piece here. It's that forward thinking, couple of steps ahead that we didn't know horses were capable of.

Chris - Obviously they didn't evolve this behaviour with you coming along after millions of years of evolution to play a game with them. They must have evolved this to give them some kind of competitive advantage in the wild or in their evolutionary time.

Louise - That's what's really interesting about this, Chris, because horses don't really have to strategise in this way. In the wild, horses generally eat grass, which is available to them all of the time. They don't need to stalk and hunt prey or search for food very often. We're not quite sure yet why horses are able to do this or what benefit it might have to them. We're really hoping that we can do some more research into this and find out that answer to your question.