A couple comments:

1. my $decimals = $ARGV[1] ? $ARGV[1] : 2; won't perform as expected if the user enters 0. You will get expected behavior if you use my $decimals = defined $ARGV[1] ? $ARGV[1] : 2;. You should also swap your $float assignment.
2. On the same point, if you have v5.10 or greater, you can use the 'defined or' operator (//)to do the same thing: my $decimals = $ARGV[1] // 2;
3. The & is largely unnecessary in modern Perl; your code functions just fine without it.
4. I'd be a little gun-shy about using the same variable name in the script level variables as in the subroutine level variables - it's easy to introduce a difficult bug.
5. You can use list assignment rather than a series of shifts if you want to save a few key strokes, replacing

     my $float = shift;
     my $dec = shift;
   [download]

   with

     my ($float, $dec) = @_;

   This is almost entirely cosmetic in this context.

I would also point out your algorithm will still be subject to the whims of these small deviations between string and double precision representation. Consider the default case of:

#!/usr/bin/perl
use strict;
use warnings;

my $number = defined $ARGV[0] ? $ARGV[0] : 1.2549999999999999;
my $places = defined $ARGV[1] ? $ARGV[1] : 2;

printf "%.20f\n", $number; 
print round( $number , $places ), "\n";


sub round {
  my ($float, $decimals) = @_;
  
  my $int_leftShiftFloat = int( $float * 10**($decimals + 1) );
  my $int_Round = int( ( $int_leftShiftFloat + 5 ) / 10 );
  my $float_rightShiftInt = $int_Round / 10**$decimals;
  my $float_Result = $float_rightShiftInt;
  
  return $float_Result
}
[download]

which outputs:

1.25499999999999990000
1.26
[download]