Average expert’s payoff over 1000 simulations against different DMs. The table is split into five sections, from top to bottom: Our model (AE), static rules, dynamic rules, algorithms, and the results in the numerical communication setup, which are not directly comparable to the above, text-based communication results. For each condition, we report the average expert payoff over our 1000 simulations, as well as 95% CI (in brackets, using bootstrap re-sampling with 1000 re-samples of our original 1000 simulations; see Dror et al., 2018). The human experts in the experiments ofApel et al. (2020) achieve an average payoff of 7.36.
| Expert/DM . | HC-LSTM . | BERT-LSTM . | HC-LSTM+0.1 . | HC-LSTM+0.2 . | HC-LSTM-0.1 . | HC-LSTM-0.2 . | AVG . |
|---|---|---|---|---|---|---|---|
| AE | 7.12 [7.02, 7.22] | 7.04 [7.03, 7.29] | 8.10 [8.02, 8.19] | 8.77 [8.70, 8.84] | 6.04 [5.93, 6.20] | 5.02 [4.90, 5.13] | 7.02 |
| RAND | 6.54 [6.49, 6.70] | 6.67 [6.56, 6.77] | 7.56 [7.47, 7.65] | 8.31 [8.24, 8.38] | 5.58 [5.47, 5.68] | 4.49 [4.38, 4.60] | 6.53 |
| MEDIAN | 6.46 [6.37, 6.54] | 6.85 [6.76, 6.96] | 7.24 [7.16, 7.33] | 8.02 [7.96, 8.11] | 5.45 [5.37, 5.54] | 4.66 [4.56, 4.76] | 6.45 |
| HIGHEST | 6.77 [6.65, 6.89] | 7.82 [7.73, 7.92] | 7.94 [7.84, 8.04] | 8.82 [8.74, 8.89] | 5.55 [5.42, 5.68] | 4.46 [4.33, 4.58] | 6.89 |
| EXTREMIST | 6.21 [6.11, 6.32] | 6.86 [6.76, 6.96] | 7.24 [7.14,7.34] | 7.99 [7.92, 8.09] | 5.14 [5.04, 5.26] | 4.08 [3.97, 4.19] | 6.25 |
| A-LIAR | 6.54 [6.42, 6.65] | 7.14 [7.06, 7.28] | 7.15 [7.06, 7.28] | 8.69 [8.61, 8.77] | 5.40 [5.28, 5.51] | 4.35 [4.24, 4.47] | 6.55 |
| PTD-HC | 6.88 [6.78, 6.99] | 7.03 [6.86, 7.06] | 7.68 [7.63, 7.80] | 8.49 [8.43, 8.57] | 5.83 [5.72, 5.95] | 4.92 [4.79, 5.02] | 6.83 |
| PTD-BERT | 6.79 [6.67, 6.88] | 6.59 [6.51, 6.73] | 7.72 [7.63, 7.82] | 8.46 [8.38, 8.54] | 5.77 [5.64, 5.88] | 4.82 [4.71, 4.93] | 6.69 |
| VM-SM | 6.58 [6.50, 6.71] | 7.00 [6.91, 7.12] | 7.70 [7.60, 7.77] | 8.34 [8.26, 8.41] | 5.65 [5.58, 5.79] | 4.67 [4.57, 4.80] | 6.66 |
| AE-DM2 | 7.05 [6.93, 7.14] | 7.23 [7.13, 7.33] | 7.94 [7.86, 8.02] | 8.66 [8.58, 8.73] | 5.92 [5.84, 6.07] | 4.97 [4.89, 5.10] | 6.96 |
| AE-VM2 | 7.03 [6.93, 7.13] | 7.05 [6.96, 7.17] | 8.00 [7.93, 8.09] | 8.76 [8.72, 8.85] | 5.98 [5.88, 6.09] | 4.98 [4.90, 5.12] | 6.97 |
| AE-SG | 7.53 [7.39, 7.64] | – | 8.63 [8.48, 8.65] | 9.10 [9.09, 9.22] | 6.02 [5.93, 6.23] | 4.85 [4.61, 4.93] | 7.23 |
| Expert/DM . | HC-LSTM . | BERT-LSTM . | HC-LSTM+0.1 . | HC-LSTM+0.2 . | HC-LSTM-0.1 . | HC-LSTM-0.2 . | AVG . |
|---|---|---|---|---|---|---|---|
| AE | 7.12 [7.02, 7.22] | 7.04 [7.03, 7.29] | 8.10 [8.02, 8.19] | 8.77 [8.70, 8.84] | 6.04 [5.93, 6.20] | 5.02 [4.90, 5.13] | 7.02 |
| RAND | 6.54 [6.49, 6.70] | 6.67 [6.56, 6.77] | 7.56 [7.47, 7.65] | 8.31 [8.24, 8.38] | 5.58 [5.47, 5.68] | 4.49 [4.38, 4.60] | 6.53 |
| MEDIAN | 6.46 [6.37, 6.54] | 6.85 [6.76, 6.96] | 7.24 [7.16, 7.33] | 8.02 [7.96, 8.11] | 5.45 [5.37, 5.54] | 4.66 [4.56, 4.76] | 6.45 |
| HIGHEST | 6.77 [6.65, 6.89] | 7.82 [7.73, 7.92] | 7.94 [7.84, 8.04] | 8.82 [8.74, 8.89] | 5.55 [5.42, 5.68] | 4.46 [4.33, 4.58] | 6.89 |
| EXTREMIST | 6.21 [6.11, 6.32] | 6.86 [6.76, 6.96] | 7.24 [7.14,7.34] | 7.99 [7.92, 8.09] | 5.14 [5.04, 5.26] | 4.08 [3.97, 4.19] | 6.25 |
| A-LIAR | 6.54 [6.42, 6.65] | 7.14 [7.06, 7.28] | 7.15 [7.06, 7.28] | 8.69 [8.61, 8.77] | 5.40 [5.28, 5.51] | 4.35 [4.24, 4.47] | 6.55 |
| PTD-HC | 6.88 [6.78, 6.99] | 7.03 [6.86, 7.06] | 7.68 [7.63, 7.80] | 8.49 [8.43, 8.57] | 5.83 [5.72, 5.95] | 4.92 [4.79, 5.02] | 6.83 |
| PTD-BERT | 6.79 [6.67, 6.88] | 6.59 [6.51, 6.73] | 7.72 [7.63, 7.82] | 8.46 [8.38, 8.54] | 5.77 [5.64, 5.88] | 4.82 [4.71, 4.93] | 6.69 |
| VM-SM | 6.58 [6.50, 6.71] | 7.00 [6.91, 7.12] | 7.70 [7.60, 7.77] | 8.34 [8.26, 8.41] | 5.65 [5.58, 5.79] | 4.67 [4.57, 4.80] | 6.66 |
| AE-DM2 | 7.05 [6.93, 7.14] | 7.23 [7.13, 7.33] | 7.94 [7.86, 8.02] | 8.66 [8.58, 8.73] | 5.92 [5.84, 6.07] | 4.97 [4.89, 5.10] | 6.96 |
| AE-VM2 | 7.03 [6.93, 7.13] | 7.05 [6.96, 7.17] | 8.00 [7.93, 8.09] | 8.76 [8.72, 8.85] | 5.98 [5.88, 6.09] | 4.98 [4.90, 5.12] | 6.97 |
| AE-SG | 7.53 [7.39, 7.64] | – | 8.63 [8.48, 8.65] | 9.10 [9.09, 9.22] | 6.02 [5.93, 6.23] | 4.85 [4.61, 4.93] | 7.23 |