WEBVTT 1 00:00:09.437 --> 00:00:13.205 So far, we have talked about the importance of understanding absolute 2 00:00:13.205 --> 00:00:16.760 levels of rewards and what rewards people value. 3 00:00:16.760 --> 00:00:20.730 We also discussed the importance of understanding relative rewards and 4 00:00:20.730 --> 00:00:23.680 the notion of equity where we socially compare to other peers. 5 00:00:23.680 --> 00:00:29.850 I would like to expand our discussion by talking about schedules of reinforcement, 6 00:00:29.850 --> 00:00:33.719 which is when and how a given reward is distributed. 7 00:00:35.490 --> 00:00:37.020 Consider the following problem. 8 00:00:37.020 --> 00:00:38.910 You're trying to motivate your daughter to do her homework. 9 00:00:40.290 --> 00:00:43.730 Which schedule of reinforcement would most effectively motivate your child to do her 10 00:00:43.730 --> 00:00:46.330 homework and keep doing her homework in the future? 11 00:00:47.510 --> 00:00:50.940 Option A, you give your daughter a piece of her favorite candy after 12 00:00:50.940 --> 00:00:53.110 every 20 minutes of studying. 13 00:00:53.110 --> 00:00:57.090 Option B, you give your daughter three pieces of her favorite candy after sixty 14 00:00:57.090 --> 00:00:58.710 minutes of studying. 15 00:00:58.710 --> 00:01:02.110 Option C, you give your daughter two pieces of candy after the first thirty 16 00:01:02.110 --> 00:01:06.060 minutes of studying, no candy after the next thirty minutes and 17 00:01:06.060 --> 00:01:08.889 then one piece of candy after every fifteen minutes of studying. 18 00:01:10.720 --> 00:01:11.367 What would you choose? 19 00:01:14.067 --> 00:01:18.110 Now let's assume that your daughter's willing to study for about two hours. 20 00:01:18.110 --> 00:01:21.890 So the absolute level of reward distributed is exactly the same. 21 00:01:23.270 --> 00:01:26.110 What varies is when and how its given out. 22 00:01:27.250 --> 00:01:30.080 So we'll come back to this question shortly, but 23 00:01:30.080 --> 00:01:34.700 I would like to first of all say that the most prevalent schedules of reinforcement 24 00:01:34.700 --> 00:01:38.050 in contemporary organizations are fixed interval and fixed ratio. 25 00:01:39.150 --> 00:01:43.380 By fixed interval I mean you receive a reward after a fixed time interval 26 00:01:43.380 --> 00:01:45.448 receiving a paycheck at the end of the month or biweekly. 27 00:01:45.448 --> 00:01:50.570 A fixed ratio is when you receive a reward after a fixed number of responses, 28 00:01:50.570 --> 00:01:52.830 such as you get paid after every five cars sold. 29 00:01:54.850 --> 00:01:58.200 And even though these are the most prevalent schedules of reinforcement, 30 00:01:58.200 --> 00:01:59.755 they're not necessarily the most effective. 31 00:01:59.755 --> 00:02:04.764 And In the words of B.F. Skinner, no one works on Monday morning 32 00:02:04.764 --> 00:02:09.420 because he is reinforced by a paycheck on Friday afternoon. 33 00:02:09.420 --> 00:02:13.326 I'm gonna show you a video that will give us insight into a vastly different 34 00:02:13.326 --> 00:02:14.880 schedule of reinforcement. 35 00:02:15.970 --> 00:02:16.514 Take a look. 36 00:02:24.702 --> 00:02:26.782 One of the reasons slot machines are so 37 00:02:26.782 --> 00:02:31.540 addictive is because they're based on the variable schedule of reinforcement. 38 00:02:31.540 --> 00:02:35.320 Specifically variable ratio schedule of reinforcement. 39 00:02:35.320 --> 00:02:38.610 What I mean by this is that the probability is constant. 40 00:02:38.610 --> 00:02:41.350 But the number of lever presses needed to win is variable. 41 00:02:43.000 --> 00:02:46.430 So again, the variable ratio schedule of reinforcement 42 00:02:46.430 --> 00:02:50.170 is the number of units produced to receiver award various. 43 00:02:50.170 --> 00:02:52.340 An example would be a lottery for your employees. 44 00:02:52.340 --> 00:02:53.980 Introducing probabilistic rewards. 45 00:02:53.980 --> 00:02:55.450 I'll give an example of those shortly. 46 00:02:57.000 --> 00:03:00.340 There's also variable interval schedule reinforcement, 47 00:03:00.340 --> 00:03:04.020 where you receive reward after timed intervals of different length. 48 00:03:04.020 --> 00:03:05.980 An example would be receiving praise only now and 49 00:03:05.980 --> 00:03:07.980 then, a surprise inspection, a pop quiz. 50 00:03:09.510 --> 00:03:11.830 So what we know from research is the following. 51 00:03:11.830 --> 00:03:15.090 First of all, ratio reinforcement schedules 52 00:03:15.090 --> 00:03:18.100 typically outperform interval reinforcement schedules. 53 00:03:20.660 --> 00:03:25.220 And secondly is variable reinforcement schedules typically outperform their 54 00:03:25.220 --> 00:03:26.760 fixed counterparts. 55 00:03:26.760 --> 00:03:32.100 So, variable interval schedule outperforms fixed interval schedule. 56 00:03:32.100 --> 00:03:35.370 The variable ration schedule outperforms fixed ration schedule. 57 00:03:36.940 --> 00:03:40.790 When I say outperforms, it means it leads to high levels of motivation, engagement, 58 00:03:40.790 --> 00:03:41.550 and performance. 59 00:03:41.550 --> 00:03:47.190 Let me give you an example of a study, that directly compares the performance 60 00:03:47.190 --> 00:03:52.190 effects of a fixed ratio reinforcement schedule, relative to a variable ratio. 61 00:03:53.940 --> 00:03:56.990 So in this study participants were asked to do a simple task, 62 00:03:56.990 --> 00:03:58.840 which was grading exams. 63 00:03:58.840 --> 00:04:04.591 And for the first week all they received was fixed compensation, so $1.5 per hour. 64 00:04:05.700 --> 00:04:10.430 In the second week, that's where it gets interesting, bonuses were introduced. 65 00:04:10.430 --> 00:04:14.445 For the first group the bonus was as follows, 66 00:04:14.445 --> 00:04:19.840 $0.50 for each exam sheet graded if you correctly guess a coin flip. 67 00:04:19.840 --> 00:04:21.930 So you create an exam sheet, you submit it, 68 00:04:21.930 --> 00:04:23.960 and you have to guess a coin flip correctly. 69 00:04:23.960 --> 00:04:25.850 If you guess it correctly, you get $0.50. 70 00:04:25.850 --> 00:04:27.800 If you don't guess it, you receive nothing. 71 00:04:29.670 --> 00:04:33.653 In the second condition in the second group, you get $0.25 guaranteed for 72 00:04:33.653 --> 00:04:35.710 every exam sheet graded. 73 00:04:35.710 --> 00:04:36.720 No coin flip involved. 74 00:04:37.840 --> 00:04:40.340 So in the first group, you can see this is a probabilistic reward. 75 00:04:40.340 --> 00:04:46.900 Now the first thing to recognize here, is that the absolute levels of reward, 76 00:04:46.900 --> 00:04:50.350 of compensation, are very much comparable across the two groups, 77 00:04:50.350 --> 00:04:53.500 assuming that you can guess coin flips with about 50% probability. 78 00:04:53.500 --> 00:04:55.910 And that's exactly what happened here. 79 00:04:55.910 --> 00:04:58.740 So most people guessed coin flips with about a 50% probability. 80 00:04:59.980 --> 00:05:04.880 The only thing that's different across the two bonuses, is that 81 00:05:04.880 --> 00:05:09.390 we have a variable ratio reinforcement schedule in the first condition, and 82 00:05:09.390 --> 00:05:11.150 fixed ratio in the second condition. 83 00:05:12.850 --> 00:05:17.620 I call the variable ratio reinforcement schedule In the first condition, 84 00:05:17.620 --> 00:05:21.910 because I can guess the coin flip correctly on the first exam sheet graded, 85 00:05:21.910 --> 00:05:25.400 then miss it on the second and third, then guess it again correctly on the fourth, 86 00:05:25.400 --> 00:05:28.490 and miss it again on the fifth, and guess it correctly on the sixth. 87 00:05:28.490 --> 00:05:35.150 So the rewards were becoming after different numbers of units graded. 88 00:05:35.150 --> 00:05:36.790 Let's look at the performance results. 89 00:05:36.790 --> 00:05:41.840 So what this graph shows on the y axis, is the increase 90 00:05:41.840 --> 00:05:46.520 in the number of exams graded per day following the introduction of bonuses. 91 00:05:48.190 --> 00:05:51.330 And as you can see, for the fixed-ratio reinforcement schedule, 92 00:05:51.330 --> 00:05:56.320 where people had $0.25 per exemption guaranteed, no probabilities involved, 93 00:05:56.320 --> 00:05:57.720 no coin flip, no uncertainty. 94 00:05:58.780 --> 00:06:02.158 The productivity went up by about 36.5%. 95 00:06:02.158 --> 00:06:07.080 And in the second condition, for the second group, which had the variable 96 00:06:07.080 --> 00:06:12.224 ratio of schedule reinforcement, the productivity went up by 44.8%. 97 00:06:14.690 --> 00:06:17.380 So, you can see that the schedule of reinforcement matters. 98 00:06:20.250 --> 00:06:26.170 Variable ratio schedule reinforcement can increase productivity and engagement. 99 00:06:26.170 --> 00:06:26.820 So keep that in mind. 100 00:06:26.820 --> 00:06:28.890 Let me give you another example. 101 00:06:28.890 --> 00:06:31.360 New York Life Insurance. 102 00:06:31.360 --> 00:06:32.940 What they start is, 103 00:06:32.940 --> 00:06:37.290 employees with perfect attendance are entered into lottery system. 104 00:06:37.290 --> 00:06:40.380 The odds of winning are extremely low, and people compete for 105 00:06:40.380 --> 00:06:44.530 such prizes as small cash prizes or extra days of vacation. 106 00:06:44.530 --> 00:06:47.770 But what they found is a stunning effect of reducing absenteeism 107 00:06:47.770 --> 00:06:50.070 throughout the entire organization by 21%. 108 00:06:50.070 --> 00:06:55.435 [SOUND] How many of you have just looked at your phones? 109 00:06:55.435 --> 00:07:00.120 Now our phones is a great a example of a variable reinforcement schedule. 110 00:07:00.120 --> 00:07:03.510 And just like with gambling, we sometimes get addicted to or phones and 111 00:07:03.510 --> 00:07:06.630 iWatches, and hear this phantom buzzes and rings. 112 00:07:08.530 --> 00:07:11.710 Now coming back to the question I posted for you in the beginning of the session. 113 00:07:12.840 --> 00:07:16.430 I'm sure you recognize by now that options A and 114 00:07:16.430 --> 00:07:20.950 B are fixed interval reinforcement schedules. 115 00:07:20.950 --> 00:07:24.800 Option C is a variable interval reinforcement schedule. 116 00:07:24.800 --> 00:07:28.350 And so, out of these three choices, you might consider giving option C a shot. 117 00:07:28.350 --> 00:07:32.530 Because on average tends to outperform fixed interval reinforcement schedules. 118 00:07:33.720 --> 00:07:36.800 Now if you want to reward your daughter for the work actually done and not for 119 00:07:36.800 --> 00:07:39.900 the time spent studying, think of the folly 120 00:07:39.900 --> 00:07:43.210 of rewarding one thing while expecting else that we just discussed. 121 00:07:43.210 --> 00:07:47.410 You may consider a variable-ratio reinforcement schedule. 122 00:07:47.410 --> 00:07:51.280 That would look like giving her a piece of candy for the first problem solved, 123 00:07:51.280 --> 00:07:54.930 another piece of candy for the second problem solved, no piece of candy for 124 00:07:54.930 --> 00:07:57.000 the third, maybe four pieces of candy for the fourth. 125 00:08:00.560 --> 00:08:04.470 So, to reflect on the key insights from this discussion so far. 126 00:08:05.620 --> 00:08:09.370 I think the most important thing to recognize is that in addition to attaining 127 00:08:09.370 --> 00:08:13.960 to absolute levels of rewards, and what rewards people value, and 128 00:08:13.960 --> 00:08:18.370 to relative levels of rewards, as we discussed in the equity conversation. 129 00:08:19.450 --> 00:08:22.990 It's really important to understand that the schedule of reinforcement, when and 130 00:08:22.990 --> 00:08:25.850 how a given reward is distributed matters greatly. 131 00:08:26.880 --> 00:08:32.480 What we know from research is that ratio schedules typically are more effective 132 00:08:32.480 --> 00:08:33.970 than interval schedules. 133 00:08:33.970 --> 00:08:37.420 But what's even more insightful is that fixed interval and 134 00:08:37.420 --> 00:08:42.290 fixed ratio schedules even though they're the most prevalent in organizations 135 00:08:42.290 --> 00:08:44.490 doesn't mean they're the most effective. 136 00:08:44.490 --> 00:08:49.300 Consider using variable interval over fixed interval reinforcement schedule. 137 00:08:49.300 --> 00:08:51.960 And variable ratio over fixed ratio reinforcement schedule. 138 00:08:53.570 --> 00:08:56.470 And also think about the fact that incentives are pervasive. 139 00:08:58.330 --> 00:09:02.260 Think about how you can use incentive to improve not just your work life, but 140 00:09:02.260 --> 00:09:03.240 also life outside of work.