Solvay Conference

This is what it looks like to change the world.

We posted previously about FACE++ and the article from Theresa Kuntzler showing how to integrate this into R. I couldn’t resist taking this for a spin one more time, on this occasion using it to examine one of the most famous conference photographs of them all, the one from the 5th Solvay Conference of 1927.

Here is the photograph, it features a veritable who’s-who of physicists and chemists from a century ago. As a physicist, seeing the people that gave rise to so many brilliant theories and techniques always brings a thrill.

.

We ran the image through FACE++ to generate a tibble, called faces, of names and emotions. Here it is:

faces %>% 
  select(scientist, anger:surprise) %>% 
  kable("html") %>%
  kable_styling() %>%
   scroll_box(
    height = "200px",
    box_css = "border: 1px solid #ddd; padding: 5px; ",
    fixed_thead = TRUE
  )

scientist	anger	disgust	fear	happiness	neutral	sadness	surprise
compton	0.000	0.104	0.000	0.004	99.891	0.000	0.000
langevin	27.209	4.816	0.295	0.804	46.320	9.604	10.952
debroglie	0.151	0.092	11.442	0.092	27.725	38.821	21.677
guye	0.147	0.147	4.745	1.453	63.292	30.049	0.167
born	0.005	0.006	0.005	4.869	94.881	0.173	0.059
curie	0.000	0.000	0.000	0.017	99.982	0.000	0.000
kramers	0.026	0.014	3.978	0.024	95.930	0.014	0.014
lorentz	97.117	0.009	0.009	0.009	0.027	2.678	0.151
dirac	0.000	0.000	0.002	0.020	98.913	0.000	1.064
einstein	0.001	0.001	0.001	0.001	99.922	0.001	0.075
debye	67.131	0.087	0.182	0.009	32.355	0.036	0.199
langmuir	0.008	0.008	0.097	0.105	99.332	0.008	0.442
knudsen	0.000	0.000	0.000	0.000	100.000	0.000	0.000
planck	89.112	0.014	0.014	0.148	10.300	0.026	0.385
bragg	0.000	0.000	0.000	0.000	100.000	0.000	0.000
heisenberg	0.000	0.000	0.000	99.996	0.002	0.000	0.002
fowler	0.002	0.000	0.004	0.001	99.948	0.044	0.000
brillioun	0.001	0.000	0.006	0.000	99.991	0.000	0.002
piccard	30.095	0.018	0.056	0.018	69.260	0.202	0.349
henriot	0.002	0.002	0.027	0.005	99.938	0.014	0.012
ehrenfest	7.918	0.081	0.336	0.949	90.683	0.017	0.017
herzen	0.077	0.865	0.077	32.781	66.045	0.077	0.077
dedonder	0.012	0.012	0.053	12.241	86.798	0.012	0.872
schroddinger	0.013	0.000	0.000	0.001	99.985	0.000	0.001
vershaffelt	3.014	0.069	2.943	0.761	90.604	0.069	2.540
pauli	0.000	0.000	0.000	0.000	99.838	0.001	0.161
wilson	1.850	10.580	0.160	22.290	63.527	0.160	1.433
bohr	0.239	0.049	0.063	0.049	69.860	29.690	0.049
richardson	0.001	0.000	0.000	0.000	99.994	0.000	0.004

As you can see, the over-riding emotion comes out as being neutral. The modern vogue of having a sea of smiling faces for your conference photograph wasn’t obviously a thing back in 1927. But there is enough variation to warrant taking a closer look. Below we show boxplots of all seven emotions. The percentage % is transformed to a logit scale to highlight variations. From the outliers we can see signs of sadness and disgust and that someone is particularly happy.

my_colours <- paletteer::paletteer_d("yarrr::eternal", n = 7)

box <- faces %>% 
  select(scientist, anger:surprise) %>% 
  pivot_longer(-scientist, names_to = "emotion", values_to = "percentage") %>% 
  ggplot(aes(emotion, percentage %>% gtools::logit(max = 100))) +
  geom_boxplot(aes(fill = emotion), show.legend = F) +
  scale_fill_manual(values = my_colours) +
  labs(x = "", y = "Percentage %") +
  theme(axis.text.x = element_blank(),
        axis.text.y = element_text(color = my_colours)) +
  coord_flip()

box_dat <- ggplot_build(box)$data[[1]]

box + 
  geom_segment(data = box_dat, 
               aes(x=xmin, xend=xmax,
                   y=middle, yend=middle), 
               colour="grey80", size=1)

Heisenberg - Happiness 100 %

Lorentz - Anger 97.1 %

de Broglie - Sadness 38.8 %

It seems that FACE++ is remarkably good at picking up pretty subtle facial expressions, even in a photograph of 29 people from 94 years ago. Not bad.