RCI07
Molly Lewis
2016-01-03
1 Pre-process from database
setwd('/Documents/GRADUATE_SCHOOL/Projects/RCI/RCI_07/RCI_07_experiment/')
db_name = "RCI07_Run1.db"
table_name = "RCI07"
sqlite <- RSQLite::SQLite()
exampledb <- dbConnect(sqlite, db_name)
db_query = dbGetQuery(exampledb, paste("SELECT datastring FROM ",
table_name, " WHERE status = 4 AND extraG1 = 0", sep = ""))
D = dejsonify(db_query)
# drop weirdo columns
drops <- c("templates","template", "action")
D = D[,!(names(D) %in% drops)]
# change column data types
factor_cols <- names(D)[c(1,4,6,11,12,14,15, 17,18:20)]
numeric_cols <- names(D)[c(13)]
D[factor_cols] <- lapply(D[factor_cols], as.factor)
D[numeric_cols] <- lapply(D[numeric_cols], as.numeric)
d = D[D$phase == 'TEST',]2 Exclusions
#multiword responses
d$guessed_label_Nwords = sapply(gregexpr("\\S+", d$guessedLabel), length)
d = d[d$guessed_label_Nwords == 1,]3 Analysis #1: Forms in the lexicon
3.1 Length
d.G0 = d %>%
mutate(guessedLabelLen = nchar(as.character(word)), gen = 0) %>%
select(guessedLabelLen, gen, condition)
length.ms <- d %>%
filter(block == 1) %>%
select(guessedLabelLen, gen, condition) %>%
bind_rows(d.G0) %>%
group_by(gen, condition) %>%
multi_boot_standard(column="guessedLabelLen")
ggplot(length.ms, aes(x = gen, y = mean, color = condition)) +
geom_pointrange(aes(ymax = ci_upper, ymin = ci_lower),size = 1.2) +
geom_line() +
scale_x_continuous(breaks=seq(0, numGen, 1)) +
ylim(1,8) +
ylab("Mean length") +
xlab("Generation") +
ggtitle("Mean length") +
themeML
3.2 Unique
unique.ms <- d %>%
filter(block == 2) %>%
group_by(gen, chain, condition) %>%
summarise(numUnique = length(unique(guessedLabel))) %>%
group_by(gen, condition) %>%
multi_boot_standard(column="numUnique")
ggplot(unique.ms, aes(x = gen, y = mean, group = condition, color = condition)) +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower), size = 1.2) +
geom_line() +
ylab("Number of unique words") +
geom_abline(intercept = 10, slope = 0, color = "grey", lty = 2) +
scale_y_continuous(limits=c(1, 14), breaks=seq(1, 12, 2)) +
xlab("Generation") +
ggtitle("Number of unique words") +
scale_x_continuous(breaks=seq(0, numGen, 1)) +
themeML
3.3 Accuracy
d$accuracy = as.factor(ifelse(as.character(d$word) == as.character(d$guessedLabel),
"correct", "incorrect"))
accuracy.ms = d %>%
group_by(gen, condition, block, chain) %>%
summarize(p_correct = sum(accuracy=="correct")/length(accuracy)) %>%
group_by(gen, condition, block) %>%
summarize(p_correct = mean(p_correct))
ggplot(accuracy.ms, aes(y = p_correct, color = condition, x = gen)) +
geom_point(size = 2) +
geom_line() +
scale_x_continuous(breaks=seq(0, numGen, 1)) +
facet_grid(.~ block) +
ylab("Proportion correct") +
xlab("Condition") +
ylim(0,1)+
ggtitle("Mean accuracy") +
themeML
3.4 CV ratio
getCVRatio <- function (d) {
word = as.character(d)
consonants = c("b", "c", "d", "f", "g", "h", "j",
"k","l", "m", "n", "p", "q", "r", "s", "t", "v", "w", "x", "y", "z")
vowels = c("a", "e", "i", "o", "u")
letters = unlist(strsplit(word, ""))
num_vowels = length(intersect(vowels, letters))
num_consonants = length(intersect(consonants, letters))
ratio = num_consonants/num_vowels
return(ratio)
}
d$CVratio = unlist(lapply(d$guessedLabel, function(d) {getCVRatio(d)}))
d$CVratio = unlist(lapply(d$CVratio, function(x) replace(x, is.infinite(x),NA)))
cv.ms = d %>%
group_by(gen, condition) %>%
multi_boot_standard(column="CVratio", na.rm = T)
ggplot(cv.ms, aes(x = gen, y = mean, color = condition)) +
geom_line() +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower), size = 1.2) +
ylab("CV ratio") +
xlab("Generation") +
scale_x_continuous(breaks=seq(0, numGen, 1)) +
themeML
3.5 Levenshtein Distance
Normalized
d$lev = NA
d$ins = NA
d$del = NA
d$sub = NA
for (i in 1:dim(d)[1]){
if (d$guessed_label_Nwords[i] == 1) { # deal with multiword cases
d$lev[i] = adist(d$word[i], d$guessedLabel[i])
d$ins[i] = drop(attr(adist(d$word[i], d$guessedLabel[i], counts = TRUE), "counts"))[1]
d$del[i] = drop(attr(adist(d$word[i], d$guessedLabel[i], counts = TRUE), "counts"))[2]
d$sub[i] = drop(attr(adist(d$word[i], d$guessedLabel[i], counts = TRUE), "counts"))[3]
}
}
d$labelLen = nchar(as.character(d$word))
# normalize Levenshetein distance by length of longer word
d$long_length = ifelse(d$labelLen > d$guessedLabelLen,
d$labelLen, d$guessedLabelLen)
d$lev.n = d$lev / d$long_length
d$ins.n = d$ins / d$long_length
d$del.n = d$del / d$long_length
d$sub.n = d$sub / d$long_length
lev.ms = d %>%
group_by(gen, condition, block) %>%
multi_boot_standard(column="lev.n")
ggplot(lev.ms, aes(y=mean, x=gen,color=condition)) +
geom_line() +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower)) +
scale_size(range=c(0.5, 1.6)) +
facet_grid(.~block) +
xlab("Generation") +
ylab("Normalized edit distance") +
ggtitle("Edit distances") +
scale_x_continuous(breaks = seq(0, numGen, 1)) +
#guides(color = guide_legend(title = "Edit metric"), size=FALSE) +
themeML
ins.ms = d %>%
group_by(gen, condition, block) %>%
multi_boot_standard(column="ins.n")
ggplot(ins.ms, aes(y=mean, x=gen,color=condition)) +
geom_line() +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower)) +
scale_size(range=c(0.5, 1.6)) +
facet_grid(.~block) +
xlab("Generation") +
ylab("Normalized edit distance") +
ggtitle("Insertions") +
scale_x_continuous(breaks = seq(0, numGen, 1)) +
#guides(color = guide_legend(title = "Edit metric"), size=FALSE) +
themeML
del.ms = d %>%
group_by(gen, condition, block) %>%
multi_boot_standard(column="del.n")
ggplot(del.ms, aes(y=mean, x=gen,color=condition)) +
geom_line() +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower)) +
scale_size(range=c(0.5, 1.6)) +
facet_grid(.~block) +
xlab("Generation") +
ylab("Normalized edit distance") +
ggtitle("Deletions") +
scale_x_continuous(breaks = seq(0, numGen, 1)) +
#guides(color = guide_legend(title = "Edit metric"), size=FALSE) +
themeML
sub.ms = d %>%
group_by(gen, condition, block) %>%
multi_boot_standard(column="sub.n")
ggplot(sub.ms, aes(y=mean, x=gen,color=condition)) +
geom_line() +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower)) +
scale_size(range=c(0.5, 1.6)) +
facet_grid(.~block) +
xlab("Generation") +
ylab("Normalized edit distance") +
ggtitle("Substitutions") +
scale_x_continuous(breaks = seq(0, numGen, 1)) +
#guides(color = guide_legend(title = "Edit metric"), size=FALSE) +
themeML
4 Analysis #2: Complexity bias
d$removedCharactersC = nchar(as.character(d$g0Word)) - d$guessedLabelLen
# merge in norms
# get norms
setwd('/Documents/GRADUATE_SCHOOL/Projects/ref_complex/Papers/RC/data/norms/')
rto_norms = read.csv("rtNormsObjs_BYITEM_exp8b.csv")
co_norms = read.csv("complicatedNormsObjs_BYITEM_exp4.csv")
# add complexity norms
# complicated
index <- match(d$obj, co_norms$ratingNum)
d$c.norms <- co_norms$meanRating[index]
# rt
index <- match(d$obj, rto_norms$Answer.train_image)
d$rt.norms <- rto_norms$log.rt[index]4.1 Cumulative characters removed
cb.ms = d %>%
group_by(quintile, gen, condition) %>%
multi_boot_standard(column="removedCharactersC", na.rm = T)
ggplot(cb.ms, aes(x=quintile, y=removedCharactersC, color = condition)) +
geom_smooth(method = "lm", aes(x=quintile, y = removedCharactersC, color = condition),
data = d,position = "identity", size = 1.5, alpha = .2) +
facet_grid( ~ gen) +
xlab("Complexity norms") +
ylab("Cumulative characters removed") +
ggtitle("Complexity bias across blocks") +
themeML
4.2 Guessed label length
ggplot(d, aes(x=c.norms, y=guessedLabelLen, color = condition)) +
geom_smooth(method = "lm", aes(x=c.norms, y=guessedLabelLen), data = d,
position = "identity", size = 1.5, alpha = .2) +
#geom_point(size = 3) +
facet_grid( ~ gen ) +
xlab("Complexity norms") +
ylab("Guessed label length (chars.)") +
ggtitle("Complexity bias across blocks") +
themeML
#Guessed label length
bias = d %>%
group_by(gen, condition) %>%
summarize(cb = cor(c.norms, guessedLabelLen)) %>%
ungroup()5 Analysis #3: Form features by chain
5.1 Length
length.ms <- d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
multi_boot_standard(column="guessedLabelLen",
na.rm = T)
ggplot(length.ms, aes(x = gen, y = mean, color = condition)) +
geom_pointrange(aes(ymax = ci_upper, ymin = ci_lower), size = .6) +
facet_wrap(~chain) +
geom_line() +
scale_x_continuous(breaks=seq(1, numGen, 1)) +
ylim(1,13) +
ylab("Mean length") +
xlab("Generation") +
ggtitle("Mean length") +
themeML
length.ms <- d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(mean = mean(guessedLabelLen))
ggplot(length.ms, aes(x = mean, group = condition, fill = condition)) +
geom_density(alpha = .3) +
facet_grid(~gen) +
ggtitle("Distribution of mean length \n across chains, by generation") +
themeML
5.2 Unique
unique.ms <- d %>%
filter(block == 2) %>%
group_by(gen, chain, condition) %>%
summarise(numUnique = length(unique(guessedLabel))) %>%
group_by(gen, chain, condition) %>%
multi_boot_standard(column="numUnique")
ggplot(unique.ms, aes(x = gen, y = mean,
group = condition, color = condition)) +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower), size = .6) +
geom_line() +
facet_wrap(~chain) +
ylab("Number of unique words") +
geom_abline(intercept = 10, slope = 0, color = "grey", lty = 2) +
scale_y_continuous(limits=c(1, 11), breaks=seq(1, 12, 2)) +
xlab("Generation") +
ggtitle("Number of unique words") +
scale_x_continuous(breaks=seq(0, numGen, 1)) +
themeML
unique.ms <- d %>%
filter(block == 2) %>%
group_by(gen, chain, condition) %>%
summarise(numUnique = length(unique(guessedLabel)))
ggplot(unique.ms, aes(x = numUnique, group = condition, fill = condition)) +
geom_density(alpha = .3) +
facet_grid(~gen) +
ggtitle("Distribution of num unique \n across chains, by generation") +
themeML
5.3 Accuracy
accuracy.ms = d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(p_correct = sum(accuracy=="correct")/length(accuracy)) %>%
group_by(gen, condition, chain) %>%
summarize(p_correct = mean(p_correct))
ggplot(accuracy.ms, aes(y = p_correct, color = condition, x = gen)) +
geom_point(size = 2) +
geom_line() +
scale_x_continuous(breaks=seq(0, numGen, 1)) +
facet_wrap(~chain) +
ylab("Proportion correct") +
xlab("Condition") +
ylim(0,1)+
ggtitle("Mean accuracy") +
themeML
accuracy.ms = d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(p_correct = sum(accuracy=="correct")/length(accuracy)) %>%
group_by(gen, condition, chain) %>%
summarize(p_correct = mean(p_correct))
ggplot(accuracy.ms, aes(x = p_correct,
group = condition, fill = condition)) +
geom_density(alpha = .3) +
facet_grid(~gen) +
xlab("proportion correct") +
ggtitle("Distribution of accuracy \n across chains, by generation") +
themeML
5.4 Levenshtein Distance
lev.ms = d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
multi_boot_standard(column="lev.n")
ggplot(lev.ms, aes(y=mean, x=gen, color=condition)) +
geom_line() +
facet_wrap(~chain) +
geom_pointrange(aes(ymax = ci_upper, ymin=ci_lower)) +
scale_size(range=c(0.5, 1.6)) +
xlab("Generation") +
ylab("Normalized edit distance") +
ggtitle("Edit distances") +
scale_x_continuous(breaks = seq(0, numGen, 1)) +
themeML
lev.ms = d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(mean= mean(lev.n))
ggplot(lev.ms, aes(x = mean,
group = condition, fill = condition)) +
geom_density(alpha = .3) +
facet_grid(~gen) +
xlab("Edit distances") +
ggtitle("Distribution of edit distances \n across chains, by generation") +
themeML
6 Responses
6.1 what about?
6.1.1 no feedback
ques = D[D$phase == 'QUESTIONNAIRE',]
responses = ques %>%
group_by(condition) %>%
select(whatAbout, anythingOdd) %>%
filter(!is.na(whatAbout))
filter(responses, condition == "noFeedback") %>%
ungroup() %>%
select(whatAbout) %>%
distinct(whatAbout) %>%
print.data.frame()## whatAbout
## 1 How people can relate words to objects
## 2 To see if having two objects with the same name would be confusing.
## 3 How we form word associations? Not sure.
## 4 Episodic memory for alien words
## 5 I have no idea.
## 6 Testing memory for new words
## 7 I have no idea
## 8 memory and cognition
## 9 I don't know
## 10 testing memory
## 11 Remembering words
## 12 learning names
## 13 memory
## 14 if we remember things better if we're shown it a second time.
## 15 learning language
## 16 short term memory
## 17 Ease of memorization based on similarity between words and words that have something to do with the object
## 18 Word memorization.
## 19 memory for words
## 20 Studying your memory
## 21 exposure and memory
## 22 How well people can learn other languages, or maybe if they do better the second time around after they know what will be asked of them the first time.
## 23 To see if people can retain information unfamiliar to them?
## 24 This experiment is on quick memory.
## 25 unsure
## 26 Memory
## 27 How memorization can be affected by giving people similar words
## 28 alien language
## 29 visual speech correlation
## 30 It was about learning an alien language and memory.
## 31 testing memory and ability to learn a new language
## 32 word association
## 33 Visual and textual memory
## 34 ability to learn based on technique and repition
## 35 a test of short term memory
## 36 To test a person's memory skills.
## 37 To test how well we memorize things
## 38 How well people remember different words in another language.
## 39 Difficulty in memorizing language when only certain consonants and vowels are used.
## 40 Testing people's memory
## 41 memory capability
## 42 Learning new language
## 43 I was terrible at this game.
## 44 to see if we can remember strange words
## 45 learning about people's memory as it relates to language.
## 46 interesting
## 47 Memory for language.
## 48 Memory, image recognition
## 49 I think it was trying to observe how well people can attach certain words to objects and memorize the names.
## 50 Looking into the way people use phonetics when learning a new language
## 51 Our likelihood of remembering something when presented with it twice.
## 52 memorization and classification
## 53 memory
## 54 How we learn new languages?
## 55 nonsense word are hard to memorize without meaning/context
## 56 how well peopel associate items with words
## 57 Memory recall as it pertains to remembering items in a list
## 58 I think it may have been about how well people can learn new foreign words they've never heard for objects they have never seen.
## 59 How people learn words.
## 60 info retention
## 61 language
## 62 Recalling random words in corelation to pictures
## 63 Testing my memory.
## 64 whether seeing objects again after trying to remember the first time will help improve memory
## 65 memory test
## 66 My best guess is word/picture association
## 67 Visual Memory
## 68 Language Recognition
## 69 How well people can memorize unfamiliar words. Also, some of these objects looked very strange / wrong in their appearance.
## 70 I don't know.
## 71 whether it was easier to remember short or long names maybe
## 72 Possibly how easy it is to create and implement new words and languages.
## 73 memory strength with unfamiliar words
## 74 How quickly people can learn new words and how well they do after being re-exposed to them.
## 75 Memorizing names after viewing each item twice, then a break, then viewing them twice more.
## 76 I do not know. I tried to relate the look of each object to the sound of the name.
## 77 no clue
## 78 memory in regards to foreign objects and words
## 79 memorization
## 80 Ability to memorize names associated with items
## 81 If we can remember things
## 82 seeing how much you can get someone to memorize for .50
## 83 Word memory
## 84 To better understand the strategies people use in terms of memory.
## 85 Learning styles?6.1.2 fuzzy feedback
filter(responses, condition == "fuzzyFeedback") %>%
ungroup() %>%
select(whatAbout) %>%
distinct(whatAbout) %>%
print.data.frame()## whatAbout
## 1 how i respond to feedback from a stranger
## 2 I think this experiment was about memory
## 3 LEARNING A NEW LANGUAGE
## 4 Word memory game
## 5 memory
## 6 Memory and working together
## 7 See how well we can remember
## 8 I don't know
## 9 communication
## 10 remembering names
## 11 no idea
## 12 Memory
## 13 Memory??
## 14 How well people can remember words
## 15 How well people remember words...?
## 16 memory improvements with additional trials
## 17 learning new words
## 18 Our memory?
## 19 I'm not sure.
## 20 I have no idea
## 21 My guess is that it's testing memory?
## 22 word association
## 23 no idea!
## 24 I don't know... but I'm going to google tigabox see what aliens know!
## 25 testing memory
## 26 I guess you are looking for correlates of learning a foreign language.
## 27 Unknown
## 28 i dont know
## 29 remembering nonsensical words
## 30 remember names of obkects
## 31 Trying to see if a person would be able to learn a new language collaboratively.
## 32 memorization
## 33 Ability to recall new words
## 34 memory of different words
## 35 learning new language
## 36 memory and whether we would try to remember harder if someone was relying on us for the answers
## 37 Memorization
## 38 memory of slight differences in names of objects
## 39 not sure
## 40 Whether talking to someone improves our memory about objects?
## 41 learning a new language based on memory.
## 42 Remembering words when under pressure of being watched by someone else
## 43 aliens
## 44 to tell if i can remember fake words
## 45 No clue.
## 46 I don't know
## 47 I'm not sure
## 48 Whether you can remember words or not
## 49 test memory
## 50 Memory under stress
## 51 don't know
## 52 Could I get close enough to the actual answer.
## 53 i have no idea
## 54 It seemed to be testing memory.
## 55 SHORT TERM MEMORY ?
## 56 patent names
## 57 learning words
## 58 associating pictures with language
## 59 recalling simple objects with unusual names
## 60 unsure
## 61 I'm guessing it was about learning and memory
## 62 Learning depth of different objects
## 63 language and memory
## 64 I have absolutely no clue
## 65 remembering silly words, I have no idea honestly. But Digog is about to be part of my every day vocabulary
## 66 to see how having a use for words increases the likelyhood of your brain remembering them
## 67 to see if computer could recognize words if they're close enoughly spelt
## 68 Memory and word association
## 69 words
## 70 how well people can remember new words
## 71 Study how well people can recall things.
## 72 Memorization of words through pictures
## 73 I am unsure
## 74 Names
## 75 How a person remembers words when presented with a picture6.2 anything odd?
6.2.1 no feedback
filter(responses, condition == "noFeedback") %>%
ungroup() %>%
select(anythingOdd) %>%
distinct(anythingOdd) %>%
print.data.frame()## anythingOdd
## 1 No
## 2 objects with the same name
## 3 no
## 4 nope
## 5 sometimes there was no picture for the same word??
## 6 there were two pugs. slicerdoohickey looked like something that sliced. pog and pug are similar as is kug
## 7 No.
## 8 I didn't notice anything odd or suspicious.
## 9 I suspected the gumpig had gum in it!
## 10 None
## 11 some of the alien words were repeated
## 12 Nope
## 13 N/A
## 14 2 of the items had the same name
## 15 Similar to english words
## 16 I noticed a lot of the words ended in -ir, and some of the items were odd - 'nogato' 'bagdamir' and 'nugara' in particular (if i am remembering correctly)
## 17 Yes
## 18 I thought the color choice of the objects and the names of the objects were odd.
## 19 That some of the items had the same names
## 20 Some things sounded like real words.
## 21 repetitive names
## 22 two repeated words
## 23 the way the words were spelt
## 24 well the whole thing was kinda odd...
## 25 Rhyming the nonsense words and some objects being named the same thing
## 26 Not really, the objects were pretty weird though.
## 27 some of the names were the same
## 28 I think the thing that was labeled pipagigamag was just called gigamag earlier
## 29 seemed strange
## 30 There were 2 things called 'gup'.
## 31 It seems like the names changed but I'm not sure
## 32 Some objects looked very odd / wrong.
## 33 Multiple items had the same name: gatub.
## 34 no, nothing
## 35 no, just weird
## 36 No it seems fine to me.
## 37 Not really. There were 20 items in each set...I want to say that one of the items only showed up once and another item showed up thrice, but I don't really know, honestly. This was difficult! Fun, though.
## 38 same name for some same pictures
## 39 I believe the objects and words were made up
## 40 The alien language
## 41 It was very odd
## 42 Just that some of the gup objects seemed unrelated.6.2.2 fuzzy feedback
filter(responses, condition == "fuzzyFeedback") %>%
ungroup() %>%
select(anythingOdd) %>%
distinct(anythingOdd) %>%
print.data.frame()## anythingOdd
## 1 no
## 2 No
## 3 YES, MY PARTNER KNEW NOTHING
## 4 nothing else
## 5 my partner's responses did not seem human. they seemed repetitive.
## 6 Yes, the partner
## 7 There was no other participant, only canned answers
## 8 the words and items.
## 9 working with others
## 10 My partner had the same responses
## 11 Just odd.
## 12 The other person didn't seem to be a real person.
## 13 No.
## 14 I think my partner was fake
## 15 I think the other person was a computer
## 16 I figured that the partner was fake?
## 17 The partner wasn't real.
## 18 nathaniel was annoying
## 19 nope
## 20 what the heck
## 21 Yeah, all the weird objects and the partner.
## 22 the words
## 23 It seemed pretty easy. Am in the control group.
## 24 The partner seemed artificial
## 25 Nathanial wasnt real
## 26 my partner didnt seem human
## 27 No, everything worked well.
## 28 there wasn't a partner
## 29 The other person wasn't real
## 30 odd
## 31 the words haha
## 32 None
## 33 It didn't seem like the person I was paired with was real
## 34 the 'partner'
## 35 I don't think my partner was real
## 36 nathaniel wasn't real, but I don't find that suspicious at all.
## 37 Nope
## 38 yes
## 39 no, nothing odd
## 40 don't think there was a partner
## 41 The whole thing was odd.
## 42 Nothing
## 43 The other participant seemed like he was a computer
## 44 fake partner
## 45 The partner.
## 46 I don't think Nathaniel was real
## 47 I'm sure Nathaniel wasn't real
## 48 the names
## 49 The words were strange but that's all
## 50 PARTNER'S RESPONSES
## 51 Why was there a fake person responding to my guesses?
## 52 all clear
## 53 As long as you entered one correct alien word, it would be accepted for any of the images.
## 54 a little
## 55 person wasnt real
## 56 I'm not sure I believed Nathaniel was a real person
## 57 unlikely that I was paired with a real person
## 58 The whole thing seemed odd
## 59 aside from the silly words, no
## 60 ROG GOB SPORK ANACOP
## 61 Nathaniel wasn't real!
## 62 yes - Nathaniel's responses
## 63 i think my partner was a computer
## 64 My partner didn't seem real
## 65 Yes
## 66 Nothing seemed odd.Get difference between conditions over generations - are they bimodal?
length.diffs <- d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(mean = mean(guessedLabelLen)) %>%
group_by(gen) %>%
mutate(dif = mean - lag(mean)) %>%
filter(!is.na(dif)) %>%
select(-condition, -mean) %>%
group_by(chain) %>%
summarize(diff = sum(dif))
ggplot(length.diffs, aes(x=diff)) +
geom_bar() +
themeMLaccuracy.diffs <- d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(p_correct =
sum(accuracy == "correct")/length(accuracy)) %>%
group_by(gen, condition, chain) %>%
summarize(p_correct = mean(p_correct)) %>%
group_by(gen) %>%
mutate(dif = p_correct - lag(p_correct)) %>%
filter(!is.na(dif)) %>%
select(-condition, -p_correct) %>%
group_by(chain) %>%
summarize(diff = sum(dif))
ggplot(accuracy.diffs, aes(x=diff)) +
geom_bar() +
themeMLlev.diffs = d %>%
filter(block == 2) %>%
group_by(gen, condition, chain) %>%
summarize(mean = mean(lev.n)) %>%
group_by(gen) %>%
mutate(dif = mean - lag(mean)) %>%
filter(!is.na(dif)) %>%
select(-condition, -mean) %>%
group_by(chain) %>%
summarize(diff = sum(dif))
ggplot(lev.diffs, aes(x = diff)) +
geom_bar() +
themeML