Update: While the  (?: (b.) (?{ push @two, $2 }))+ approach discussed in this reply can certainly work, I think I would prefer the two-step approach outlined in choroba's reply above for its greater readability and maintainability.

If I take the for-loop out of the equation, I get something closer to what I expect (still Strawberry 5.14.4.1):

c:\@Work\Perl\monks>perl -wMstrict -le
"my $string = 'ab1b2b3b4d';
 my @two;
 my @matches = $string =~ /(a)
                           (b.(?{ push @two, $2 if defined $2 }))+
                           (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'ab5b6b7d';
 @two = ();
 @matches = $string =~ /(a)
                        (b.(?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'ab8b9d';
 @two = ();
 @matches = $string =~ /(a)
                        (b.(?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'abxd';
 @two = ();
 @matches = $string =~ /(a)
                        (b.(?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'ad';
 @two = ();
 @matches = $string =~ /(a)
                        (b.(?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
"
'ab1b2b3b4d' -> all:(a b4 d) -> 2:(b1 b2 b3)
'ab5b6b7d' -> all:(a b7 d) -> 2:(b5 b6)
'ab8b9d' -> all:(a b9 d) -> 2:(b8)
'abxd' -> all:(a bx d) -> 2:()
'ad' -> all:() -> 2:()
[download]

c:\@Work\Perl\monks>perl -wMstrict -le
"my $string = 'ab1b2b3b4d';
 my @two;
 my @matches = $string =~ /(a)
                           (?: (b.) (?{ push @two, $2 if defined $2 })
+)+
                           (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'ab5b6b7d';
 @two = ();
 @matches = $string =~ /(a)
                        (?: (b.) (?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'ab8b9d';
 @two = ();
 @matches = $string =~ /(a)
                        (?: (b.) (?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'abxd';
 @two = ();
 @matches = $string =~ /(a)
                        (?: (b.) (?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
 ;;
 $string = 'ad';
 @two = ();
 @matches = $string =~ /(a)
                        (?: (b.) (?{ push @two, $2 if defined $2 }))+
                        (d)/x;
 print qq{'$string' -> all:(@matches) -> 2:(@two)};
"
'ab1b2b3b4d' -> all:(a b4 d) -> 2:(b1 b2 b3 b4)
'ab5b6b7d' -> all:(a b7 d) -> 2:(b5 b6 b7)
'ab8b9d' -> all:(a b9 d) -> 2:(b8 b9)
'abxd' -> all:(a bx d) -> 2:(bx)
'ad' -> all:() -> 2:()
[download]

Update: The expression
    (?: (b.) (?{ push @two, $2 if defined $2 }))+
works just as well without the statement modifier:
    (?: (b.) (?{ push @two, $2 }))+
I don't understand the use of the statement modifier: Could  $2 ever be undefined in this situation?

Without the modifier, I'm getting one more element in the array which is undefined.

($q=q:Sq=~/;[c](.)(.)/;chr(-||-|5+lengthSq)`"S|oS2"`map{chr |+ord
}map{substrSq`S_+|`|}3E|-|`7**2-3:)=~y+S|`+$1,++print+eval$q,q,a,
[download]

Please see this reply below.

---

Give a man a fish:  <%-{-{-{-<

Ok, it looks like the for-loop business is just some bug that got fixed between (my) version 5.14.4.1 and (your) version 5.20.

As to the difference between
    (b.(?{ push @two, $2 if defined $2 }))+
and
    (?: (b.) (?{ push @two, $2 }))+

It seems to me that in the first expression, the location of the  (?{ ... }) sub-expression within the second capture group means that on the first occasion that  b. can match within string, the capture group has not actually "closed" yet, and so  $2 is undefined. On each subsequent occasion, the previous capture to  $2 is what gets pushed to the array. So, the  @two array is always "one behind" in its contents if the first (undefined) value of  $2 is excluded.

In the second expression, the second capture group is always fully closed at the moment it is pushed to the  @two array, so no weirdness.