by Guangming Lang
3 min read


  • r

This is the second article of the series on how environments work in R. Recall that in the first article, we demonstrated how environment and evaluation work in R and we learned the following:

  • the default current environment (or workspace) is called the global env.
  • when you define a function in the current environment (i.e., this is another way of saying opening Rstudio, making a funtion and sending it to Console), you effectively put that function inside the global env.
  • when you call that function in the global env, a new environment is created, binding the function arguments to the values supplied. This new environment is called the environment associated with the function, and is inserted before the global env on the search path. In other words, its parent (environment) is the global env.
  • when the body of that function is evaluated, a search is made, first in that newly created environment associated with the function, and next in the global env.
  • in general, when you define a function f inside another function h, you effectively put f inside the environment associated with h, which is NOT the calling environment of h, but the newly created environment that binds the arguments of h to values supplied. When the body of f is evaluated, R searches for values first in the environment associated with f, and then in the environment associated with h, and then in the global env (assuming h is defined in the global env).

The assignment operators in R

There’re two assignment operators in R: <- and <<-. While <- changes the values of objects in the immediate environment searched, <<- changes the values of objects first encountered, starting with the parent environment of the immediate environment and traversing upwards through the parent environments until the global environment is reached. Consider the following example. The function h uses <- to only set its local y to 0. The function to_zero uses <<- to set the y declared in the global env to 0. If no y found in the global env, it will create it in the global env and bind its value to 0.

# define functions
h = function(y) y <- 0
to_zero = function(y) y <<- 0

# initialize y in global env
y = 10

# h doesn't change the global y 
## [1] 10
# to_zero sets the global y to 0
## [1] 0

State maintanence

The combination of <<- and R evaluation model allows functions to remember variable values between funtion calls. For example, consider the following function student, which appends a 3-digit number to a given student name. It takes a string name and returns a list of two functions. Because both functions are defined inside the student function, more precisely, they are defined inside the environment associated with student, and because name will also be created in the same environment when student is called, both functions of the returned list will have access to the value of name. Notice how I use <<- to update name.

student = function(name) {
        list(print_name = function() print(name),
             add_num = function() {
                     num = ceiling(runif(1, min = 100, max = 999))
                     name <<- paste0(name, num)

We can now create name cards for different students by calling student. Notice Joe’s name card is maintained separately from Jean’s. This is what we wanted.


# create badge for joe
joe = student("Joe")
## [1] "Joe267"
# create badge for jean
jean = student("Jean")
## [1] "Jean732"

If you want to see a bigger example of <<-, click here for the final article, where I used <<- to code up the mortgage payment calculator in a different fashion.