A lexical approach for identifying behavioural action sequences
Table 1
Motifs over-represented in the exploratory dataset.
A subset of motifs occur (‘Observed’ column) more often than predicted by a first-order Markov model (the ‘Expected’ column). The p-value is obtained using a likelihood ratio test. Single-length motifs are not shown (see Fig 5c). See also S1 Table. The fifth column shows the percentage of unique fish (out of 171 total) which executed the motif at least once in the most likely partitioning of the data. Note that a motif may never appear in the most likely partitioning even though it has non-zero pm (for eg. tttt below, which is partitioned into four individual ts). The rightmost column shows the fraction of the exploratory dataset tiled by each motif.