Figures & data
Figure 1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally.
![Figure 1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally.](/cms/asset/bab768ac-8f8d-44db-8e8e-f894fc74b14e/gmas_a_2217324_f0001_b.gif)
Figure 2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally.
![Figure 2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally.](/cms/asset/7c107d61-a15e-4757-9d38-9e39de66b7e2/gmas_a_2217324_f0002_oc.jpg)
Figure 3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k.
![Figure 3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k.](/cms/asset/dad85d8b-8825-41f3-831d-c4667de83380/gmas_a_2217324_f0003_b.gif)
Figure 4. Boxplots of total search costs until convergence (based on Eq. (5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64.
![Figure 4. Boxplots of total search costs until convergence (based on Eq. (5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64.](/cms/asset/1fe49327-a6d5-44d2-8038-ebb079c4abc7/gmas_a_2217324_f0004_b.gif)
Figure 5. Proportions of agents converging on high quality (HQ) object for each value of , for all values of k; 20 runs per value of
; solid lines depict mean proportions; dots are jittered for better representation.
![Figure 5. Proportions of agents converging on high quality (HQ) object for each value of ε0, for all values of k; 20 runs per value of ε0; solid lines depict mean proportions; dots are jittered for better representation.](/cms/asset/3b1a5016-3528-4e6b-9d12-5889388590f1/gmas_a_2217324_f0005_b.gif)
Figure 6. Gini coefficient values among high quality (HQ) objects in terms of numbers of agents choosing objects for each value of , for all values of k; 20 runs per value of
; data for runs that have at least one agent converging on an HQ object; solid lines depict mean Gini.
![Figure 6. Gini coefficient values among high quality (HQ) objects in terms of numbers of agents choosing objects for each value of ε0, for all values of k; 20 runs per value of ε0; data for runs that have at least one agent converging on an HQ object; solid lines depict mean Gini.](/cms/asset/9e314b64-b4c6-4a98-9b60-864117932703/gmas_a_2217324_f0006_b.gif)
Figure 7. Search costs according to Eq. (5) for different values of , for each values of k; runs that end in all agents converging on a HQ object only.
![Figure 7. Search costs according to Eq. (5) for different values of ε0, for each values of k; runs that end in all agents converging on a HQ object only.](/cms/asset/9db60bff-1d32-410b-a224-cbc758ee6e92/gmas_a_2217324_f0007_b.gif)
Figure 8. Average outcomes of learning under the common knowledge of rationality assumption; the left panel shows mean search costs, the right panel shows mean Gini coefficients. Neighborhood sizes: black for k = 2, blue for k = 4, red for k = 8, green for k = 16, yellow for k = 32, and orange for k = 64.
![Figure 8. Average outcomes of learning under the common knowledge of rationality assumption; the left panel shows mean search costs, the right panel shows mean Gini coefficients. Neighborhood sizes: black for k = 2, blue for k = 4, red for k = 8, green for k = 16, yellow for k = 32, and orange for k = 64.](/cms/asset/7091684f-97c9-4646-97d4-168ed66cb464/gmas_a_2217324_f0008_oc.jpg)
Figure 9. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs, blue lines plot means per round; the optimal learning scenario is depicted in the left-hand plot and the “most promising” common knowledge scenario in the right-hand plot.
![Figure 9. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs, blue lines plot means per round; the optimal learning scenario is depicted in the left-hand plot and the “most promising” common knowledge scenario in the right-hand plot.](/cms/asset/0003663d-2911-4df1-94bb-7c5591278690/gmas_a_2217324_f0009_oc.jpg)
Figure B1. Search costs according to Eq. (5) for different values of , for each values of k; all runs.
![Figure B1. Search costs according to Eq. (5) for different values of ε0, for each values of k; all runs.](/cms/asset/4fc51a74-5e64-4db6-9cd4-b7a031afdf13/gmas_a_2217324_uf0010_b.gif)
Figure B2. Proportions of agents choosing a HQ object at convergence for different values of , for each value of neighborhood size k; solid lines are mean proportions.
![Figure B2. Proportions of agents choosing a HQ object at convergence for different values of ε, for each value of neighborhood size k; solid lines are mean proportions.](/cms/asset/a4749af4-1fcb-454b-a2f6-e0491d8649c6/gmas_a_2217324_uf0011_b.gif)
Figure B4. Search costs according to EquationEq. (5)(5)
(5) for different values of
across all runs for each value of k.
![Figure B4. Search costs according to EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) for different values of ε across all runs for each value of k.](/cms/asset/9396ec58-6938-4a82-ace4-1368ffaecf31/gmas_a_2217324_uf0013_oc.jpg)
Figure B5. Search costs according to EquationEq. (5)(5)
(5) for different values of
for runs ending in all agents converging on a high-quality object only, for all values of k.
![Figure B5. Search costs according to EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) for different values of ε for runs ending in all agents converging on a high-quality object only, for all values of k.](/cms/asset/6eecfca6-088b-48d1-a385-ef13b832d504/gmas_a_2217324_uf0014_b.gif)
Figure B6. Box-plots of search costs in the optimal learning condition (left-hand) and the “most promising” common knowledge condition; neighborhood size k equals 2.
![Figure B6. Box-plots of search costs in the optimal learning condition (left-hand) and the “most promising” common knowledge condition; neighborhood size k equals 2.](/cms/asset/e6464c9c-299a-45b7-9a5a-d13cdc94d4ea/gmas_a_2217324_uf0015_b.gif)
Figure B7. Boxplots if Gini coefficients at convergence in the optimal learning condition (left-hand) and the “most promising” common knowledge condition; neighborhood size k equals 2.
![Figure B7. Boxplots if Gini coefficients at convergence in the optimal learning condition (left-hand) and the “most promising” common knowledge condition; neighborhood size k equals 2.](/cms/asset/a5c6d5c5-ea57-4470-92a4-83acb97aa49e/gmas_a_2217324_uf0016_b.gif)
Table C1. Parameter values for robustness checks.
Figure C.1.1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC0.
![Figure C.1.1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC0.](/cms/asset/adb5e969-34d6-4224-8aa8-f45f6958050e/gmas_a_2217324_uf0017_oc.jpg)
Figure C.1.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC0.
![Figure C.1.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC0.](/cms/asset/0d069162-7a1b-4d33-b917-f867a2f43ced/gmas_a_2217324_uf0018_b.gif)
Figure C.1.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC0.
![Figure C.1.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC0.](/cms/asset/da14ca1d-e5d8-403a-b730-800f4224b18b/gmas_a_2217324_uf0019_b.gif)
Figure C.1.4. Boxplots of total search costs until convergence (based on equation (5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC0.
![Figure C.1.4. Boxplots of total search costs until convergence (based on equation (5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC0.](/cms/asset/6d17524c-38db-4dbb-a395-b8e92794c5dd/gmas_a_2217324_uf0020_b.gif)
Figure C.2.1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC1a.
![Figure C.2.1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC1a.](/cms/asset/b1c61297-c7b3-432e-9e84-75cf95a81e98/gmas_a_2217324_uf0021_oc.jpg)
Figure C.2.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC1a.
![Figure C.2.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC1a.](/cms/asset/866f4de6-252a-492f-baf2-39549154827d/gmas_a_2217324_uf0022_b.gif)
Figure C.2.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC1a.
![Figure C.2.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC1a.](/cms/asset/a2230fa9-94a2-4935-aca2-90ac9ae0f4fb/gmas_a_2217324_uf0023_b.gif)
Figure C.2.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5)
(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC1a.
![Figure C.2.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC1a.](/cms/asset/58c8051c-0cc6-4c61-b8eb-d2f40157cc8c/gmas_a_2217324_uf0024_b.gif)
Figure C.3.1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC1b.
![Figure C.3.1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC1b.](/cms/asset/7767b474-f26b-4afb-a2d8-c1ecc468825f/gmas_a_2217324_uf0025_oc.jpg)
Figure C.3.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC1b.
![Figure C.3.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC1b.](/cms/asset/a9898a5e-eaf1-4001-8a30-11871d8df150/gmas_a_2217324_uf0026_b.gif)
Figure C.3.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC1b.
![Figure C.3.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC1b.](/cms/asset/8899424f-9ac8-4e3b-b50d-9b256984faef/gmas_a_2217324_uf0027_b.gif)
Figure C.3.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5)
(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC1b.
![Figure C.3.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC1b.](/cms/asset/b87d3fa1-b105-4af3-91ef-54b8c903de68/gmas_a_2217324_uf0028_b.gif)
Figure C.4.1. Proportions of agents staying with their object of choice given that it failed ()on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC2a.
![Figure C.4.1. Proportions of agents staying with their object of choice given that it failed ()on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC2a.](/cms/asset/8bdea160-e76a-4c64-8b0e-485bc6cc85db/gmas_a_2217324_uf0029_oc.jpg)
Figure C.4.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC2a.
![Figure C.4.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC2a.](/cms/asset/6f8e14a8-0453-43df-8165-175c26ccc9f6/gmas_a_2217324_uf0030_b.gif)
Figure C.4.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC2a.
![Figure C.4.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC2a.](/cms/asset/d6d805c6-1844-4a2b-ae7d-5cf0fd9b813f/gmas_a_2217324_uf0031_b.gif)
Figure C.4.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5)
(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC2a.
![Figure C.4.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC2a.](/cms/asset/41757d68-d3e1-4648-a227-fc16cf9f9325/gmas_a_2217324_uf0032_b.gif)
Figure C.5.1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC2b.
![Figure C.5.1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC2b.](/cms/asset/8309473c-23d9-4ed2-a7eb-90ae1ceaf6ef/gmas_a_2217324_uf0033_oc.jpg)
Figure C.5.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC2b.
![Figure C.5.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC2b.](/cms/asset/df88be47-95bf-46b8-a884-385a2523fbd9/gmas_a_2217324_uf0034_b.gif)
Figure C.5.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC2b.
![Figure C.5.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC2b.](/cms/asset/8bc27ff7-b185-419c-8722-a01fbad5d5d4/gmas_a_2217324_uf0035_b.gif)
Figure C.5.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5)
(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC2b.
![Figure C.5.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC2b.](/cms/asset/a045265a-4a6d-447b-ac1e-003ebf174bad/gmas_a_2217324_uf0001_b.gif)
Figure C.6.1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC3a.
![Figure C.6.1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC3a.](/cms/asset/660d3453-dfdc-4b8d-8c29-32b0cd937c4a/gmas_a_2217324_uf0002_b.gif)
Figure C.6.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC3a.
![Figure C.6.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC3a.](/cms/asset/93f1ef5b-2ffe-4864-843c-eed8ca58672c/gmas_a_2217324_uf0003_b.gif)
Figure C.6.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC3a.
![Figure C.6.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC3a.](/cms/asset/e03d4ee6-cc65-4dba-8440-6c3d97f2f038/gmas_a_2217324_uf0004_b.gif)
Figure C.6.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5)
(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC3a.
![Figure C.6.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC3a.](/cms/asset/e7c92223-96b9-4569-a6a6-0f9801f8f1d8/gmas_a_2217324_uf0005_b.gif)
Figure C.7.1. Proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC3b.
![Figure C.7.1. Proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; gray lines depict 100 runs for each neighborhood size k, black lines plot means per round; agents learn optimally; RC3b.](/cms/asset/391dfe53-5581-41c7-875a-bfabfc4a20a0/gmas_a_2217324_uf0006_b.gif)
Figure C.7.2. Mean proportions of agents staying with their object of choice given that it failed () on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC3b.
![Figure C.7.2. Mean proportions of agents staying with their object of choice given that it failed (ε0n) on the y-axis by round on the x-axis; 100 runs for each neighborhood size k; black = 2 neighbors, blue = 4 neighbors, red = 8 neighbors, green = 16 neighbors, yellow = 32 neighbors, orange = 64 neighbors; agents learn optimally; RC3b.](/cms/asset/91273850-1294-4aa1-a46d-c414e4b05b05/gmas_a_2217324_uf0007_b.gif)
Figure C.7.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC3b.
![Figure C.7.3. Boxplots of the distributions of Gini coefficients of the distribution of agent choices among high quality objects; Gini coefficients of 100 runs at convergence, per value of k; agents learn optimally; RC3b.](/cms/asset/639b554c-c0ff-400f-ad4c-c7624d7e4237/gmas_a_2217324_uf0008_b.gif)
Figure C.7.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5)
(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC3b.
![Figure C.7.4. Boxplots of total search costs until convergence (based on EquationEq. (5)(5) sn=1−pn−pLpH−pL(5) summed over all rounds n per run) for all values of k; y-axis running from 0 to 350 for k equal to 2, 4, and 8, from 0 to 8000 for k equal to 16 and 32, and from 0 to 20,000 for k equal to 64; agents learn optimally; RC3b.](/cms/asset/4590f616-fe55-44d8-9d2f-25ef7c4b54f2/gmas_a_2217324_uf0009_oc.jpg)