Although it is hard to find the specific instance when the “refrigerator mother” hypothesis of autism was first used, it is not difficult to find who first proposed it. As early as his 1943 paper, Leo Kanner was calling attention to what he saw as a lack of parental warmth and attachment to their autistic children. In his 1949 paper, he attributed autism to a “genuine lack of maternal warmth” and the “Refrigerator Mother” theory of autism was born.
In retrospect, it would appear that Kanner was confusing cause and effect. It is more likely that any lack of attachment he saw between the parents and their autistic children was due to the lack of social reciprocity in the children. He consistently ignored the fact that the affected children in his 1943 paper had unaffected siblings who were, presumably, exposed to the same parents and their warmth or lack of it. In a 1960 Time Magazine interview, Kanner described the mothers of autistic children as “just happening to defrost enough to produce a child.”
As instrumental as Kanner was in forming the “Refrigerator Mother” hypothesis, it was Bruno Bettleheim who gave it widespread popularity. His articles, primarily in the 1950s and 1960s, popularized the idea that autism was caused by maternal coldness toward their children—ignoring, as Kanner did, that these same mothers had other children who were not autistic. This was, without a doubt, the low ebb of professional opinion about the parents (especially the mothers) of autistic children.
http://www.autism-watch.org/causes/rm.shtml