Удвоение '\' при парсинге выражений

R> benchr::benchmark(gsub("\\D", "", "aA456ASFA"))
Benchmark summary:
Time units : microseconds 
                           expr n.eval  min lw.qu median mean up.qu  max total
 gsub("\\\\D", "", "aA456ASFA")    100 8.97  10.2   12.5   14  13.7 66.3  1400

С другой стороны:

R> cat(deparse(substitute(gsub("\\D", "", "aA456ASFA"))))
gsub("\\D", "", "aA456ASFA")

/cc @tonytonov